LAYERS OF MANETS

VELLANKI SUMANTH REDDY 09B61A05B8

Introduction

What does make routing and communication in MANET Networks dierent from routing and communication in usual LANs, WANs etc.? The main peculiarities are the absence of wires, routers and any predened infrastructure. The nodes communicate with each other over the air and this is a very unreliable medium. Because of reection, diraction, diusion and other signal propogation properties the data exchanging by nodes is frequently lost. Besides the connections between nodes can break because the nodes move and this changes then the topology of the network. Taking into consideration all these facts, its obvious that the standard approaches applicable to the usual networks with pre-existing infrastructure arent suitable in general for the ad hoc networks. In following Id like to tell a couple of words to ISO/OSI layers and explain briey how they are realized in MANET networks.Manet is a specication The goal of this report is to present and describe the Dynamic Source Routing (DSR) OPNET. process model that has been developed at NIST. In this way it will be possible for every OPNET user or developer to use this process model in his simulations. Thus the rst part is devoted to a general overview of the DSR protocol, and of the Mobile Ad Hoc Network (MANET) for which it has been developed. Then, since the process model cannot be provided without a node model, a presentation of this node model is done in the second part of the report. Finally, the third and last part of the report focuses on the DSR process model; 1

that is the description of this model comparing to its specication and the presentation of its state machine. Detailed technical information relative to the model, such as the variables, the functions, and the packet format descriptions, are available in the appendix. It is important to raise this question since the Dynamic Source Routing protocol was designed especially for this kind of network. A Mobile Ad Hoc Network, also called a MANET, is an autonomous collection of mobile nodes forming a dynamic wireless network. The administration of such a network is decentralized, i.e. each node acts both as host and router and forwards packets for nodes that are not within transmission range of each other. A MANET provides a practical way to rapidly build a decentralized communication network in areas where there is no existing infrastructure or where temporary connectivity is needed, e.g. emergency situations, disaster relief scenarios, and military applications.

Application Manet

Layer

In

Peer-to-peer (P2P) computing is a networking and distributed computing paradigm which allows the sharing of computingresources and services by direct, symmetric interaction between computers. With the advance in mobile wireless communication technology and the increasing number of mobile users, peer-to-peer computing, in both academic research and industrial development,

has recently begun to extend its scope to address problems relevant to mobile devices and wireless networks. The mobile ad hoc network (MANET) and P2P systems share key characteristics including self-organization and decentralization, and both need to solve the same fundamental problem: connectivity. Although it seems natural and attractive to deploy P2P systems over MANET due tothis common nature, the special characteristics of mobileen vironments and the diversity in wireless networks bring new challenges for research in P2P computing.

ing environment. For this reason, several research directions have focused on trying to oer some form of multicast communication at a higher level, typically via an applicationoriented middleware. Such a higher level data dissemination support is often referred to as Application Layer Multicast (ALM). Of course, ALM-based solutions to data dissemination usually rely on the lower level networking mechanism. 2.1.2 Working

2.1
2.1.1

Application Layer Multicast

Introduction

For An increasing number of Peer-to-Peer (P2P) Internet applications rely today on data dissemination as their cornerstone, e.g., audio or video streaming, multi-party games. These applications typically depend on some support for multicast communication, where peers interested in a given data stream can join a correspondingmulticast group. As a consequence, the eciency, scalability, and reliability guaranteesof these applications are tightly coupled with that of the underlying multicastmechanism. At the network level, IP Multicast oers quite good eciency and scalability but best-eort reliability only. Moreover, the deployment of IP Multicast requires all routers to be appropriately congured, which makes it quite impractical in large scale or open settings, i.e., where one does not have full control over the network2

At the network level, IP Multicast oers quite good eciency and scalability but best-eort reliability only. Moreover, the deployment of IP Multicast requires all routers to be appropriately congured, which makes it quite impractical in large scale or open settings, i.e., where one does not have full control over the networking environment. For this reason, several research directions have focused on trying to oer some form of multicast communication at a higher level, typically via an application-oriented middleware. Such a higher level data dissemination support is often referred to as Application Layer Multicast (ALM). Of course, ALM-based solutions to data dissemination usually rely on the lower level networking mechanism.namely loosely coupled large-area networks,such as the Internet. Indeed, most solutions aim at large systems with hundreds or thousands of nodes. In these contexts, protocols do not seem to signicantly favor a design for a specic source or for multi sources, in which the same overlay network is reused by any peer willing to disseminate information.While most of the

protocols surveyed were originally designed for P2P Internet environments, some could be used in MANET settings. Yet, most of them are not suitable to MANETs. At rst glance, this may look surprising due to the similarities between P2P networks and MANETs. In reality, however, MANET components are more resource-constrained then P2P Internet peers. Moreover, some P2P protocols rely on assumptions that are not feasible in MANETs (e.g., the existence of rendezvous points). With respect to overlay structure, the large majority of protocols uses physical node IDs, as opposed to logical IDs. Moreover, by far and large most protocols build tree overlays, and the few ones based on meshes use it mainly to control information, not to disseminate data. Disseminating information over a mesh would create redundant data paths and waste bandwidth in the absence of failures 2.1.3 Conclution

Most multicast protocols surveyed focus on minimizing latency and improving scalability. These goals are justied by the nature of the environment targeted bytypical applicationlevel multicast protocols, namely loosely coupled large-area networks,such as the Internet. Indeed, most solutions aim at large systems with hundreds or thousands of nodes. In these contexts, protocols do not seem to signicantly favor a design for a specic source or for multi sources, in which the same overlay network is reused by any peer willing to disseminate information.While most of the protocols surveyed were originally designed 3

for P2P Internet environments, some could be used in MANET settings. Yet, most of them are not suitable to MANETs. At rst glance, this may look surprising due to the similarities between P2P networks and MANETs. In reality, however, MANET components are more resource-constrained then P2P Internet peers. Moreover, some P2P protocols rely on assumptions that are not feasible in MANETs (e.g., the existence of rendezvous points). With respect to overlay structure, the large majority of protocols uses physical node IDs, as opposed to logical IDs. Moreover, by far and large most protocols build tree overlays, and the few ones based on meshes use it mainly to control information, not to disseminate data. Disseminating information over a mesh would create redundant data paths and waste bandwidth in the absence of failures. While this would provide higher reliability, as we will see next, protocols tend to resort to dierent mechanisms to handle peer or link failures.It is perhaps less surprising that most solutions implement a distributed overlay creation and maintenance, where the responsibility is shared among several peers, as opposed to a centralized approach in which one peer is responsible for building the overlay. One consequence of the centralized technique is that one node must have global knowledge about the system. While most multicast protocols surveyed strive to avoid global knowledge, a few adopt a hybrid approach in which some nodes must maintain a complete view of the system. This is typically the case.

2.2

Mobile File-Sharing over concern us in this article are: P2P is generally referred to the application layer, but P2P Networks

The current list of application proles either published, or in the works are: Peer-topeer (P2P) computing is a networking and distributedcomputing paradigm which allows the sharing of computing resources and services by direct, symmetric interaction between computers. With the advance in mobile wireless communication technology and the increasing number of mobile users, peerto-peer computing, in both academic research and industrial development, has recently begun to extend its scope to address problems relevant to mobile devices and wireless networks.The mobile ad hoc network (MANET) and P2P systems share key characteristics including self-organization and decentralization, and both need to solve the same fundamental problem: connectivity. Although it seems natural and attractive to deploy P2P systems over MANET due to this common nature, the special characteristics of mobile environments and the diversity in wireless networks bring new challenges for research in P2P computing. 2.2.1 introduction

MANET is generally referred to the network layer, which is a lower layer concerning network access issues. Thus, the immediate result of this layer partition reects the difference of the packet transmission methods between P2P and MANET: the P2P overlay is a unicast network with virtual broadcast consisting of numerous single unicast packets, while the MANET overlay always performs physical broadcasting. Peers in P2P overlay are usually referred to static nodethough no priori knowledge of arriving and departing is assumed, but peers in MANET are usually referred to mobile node since connections are usually constrained by physical factors like limited battery energy, bandwidth, computing power, and so forth.

2.2.2

conclusion

Currently, most P2P systems work on wired Internet, which depends on application layer connections among peers, forming an application layer overlay network. In MANET, overlay is also formed dynamically via connections among peers, but without requiring any wired infrastructure. So the major dierences between P2P and MANET that 4

In this article, we studied the peer-to-peer systems over mobile ad hoc networks with a comparison of dierent settings for the peerto-peer overlay and underlying mobile ad hoc network. We show that the cross-layer approach performs better than separating the overlay from the access networks in Table 1. Our results would potentially provide useful guidelines for mobile operators, value-added service providers, and application developers to design and dimensionmobile peer-to-peer systems.

2.3
2.3.1

Security issues, challenges whereas nodes that not in the direct communication range use intermediate node(s) solution in MANET
abstract

Security has become a primary concern in order to provide protected communication between mobile nodes in a hostile environment. Unlike the wireline networks, the unique characteristics of mobile ad hoc networks pose a number of nontrivial challenges to security design, such as open peer-topeer network architecture, shared wireless medium, stringentresource constraints, and highly dynamic network topology. Theprovision of security services in the MANET context faces a set of challenges specic to this new technology. In this paper, we discuss security issues, vulnerable nature of the mobile ad hocnetwork, security criteria and the main attack types that exist in it. Finally we survey the current security solutions for themobile ad hoc network and then conclude this paper 2.3.2 introduction

to communicate with each other. In these two situations, all the nodes that have participated in the communication automatically form a wireless network, therefore this kind of wireless network can be viewed as mobile ad hoc network. Security has become a primary concern to provide protected communication between mobile nodes in a hostile environment. Unlike wire line networks, the unique characteristics ofmobile ad hoc networks pose a number of non-trivial challengesto the security design.

2.4
2.4.1

Vulnerabilities of MANET
Unreliability of wireless links between nodes

Because of the limited energy supply for the wireless nodes and the mobility of the nodes, the wireless links between mobile nodes in the ad hoc network are not consistent for the communication participants. 2.4.2 Dynamic topologies

In recent years mobile ad hoc networks have received tremendous attention because of their self-conguration and self-maintenance capabilities. A Mobile Ad hoc network is a system of wireless mobile nodes that dynamically selforganize in arbitrary and temporary network topologies. People and vehicles can thus be internetworked in areas without a preexisting communication infrastructure or when the use of such infrastructure requires wireless extension. In the mobile ad hoc network, nodes can directly communicate with all the other nodes within their radio ranges; 5

Nodes are free to move arbitrarily; thus, the network topology which is typically multihopmay change randomly and rapidly at unpredictable time. Because the topology of the ad hoc networks is changing constantly, it is necessary for each pair of adjacent nodes to incorporate in the routing issue so as to prevent some kind of potential attacks that try to make use of vulnerabilities in the statically cong.d routing protocol.

2.4.3

Lack of Secure Boundaries

The meaning of this vulnerability is selfevident: there is not such a clear secure boundary in the mobile ad hoc network, which can be compared with the clear line of defense in the traditional wired network. This vulnerability originates from the nature of the mobile ad hoc network: freedom to join, leave and move inside the network. Lack of secure boundaries makes the mobile ad hoc network susceptible to the attacks. Due to this mobile ad hoc network suers from allweather attacks, which can come from any node that is in the radio range of any node in the network, at any time, and target to any other node in the network. To make matters worse, there are various link attacks that can jeopardize the mobile ad hoc network, which make it even harder for the nodes in the network to resist the attacks. The attacks mainly include passive eavesdropping, active interfering, and leakage of secret information, data tampering, message replay, message contamination, and denial of service. 2.4.4

change its attack target and perform malicious behavior to dierent node in the network, thus it is very dicult to track the malicious behavior performed by a compromised node especially in a large scale ad hoc network. Therefore, threats from compromised nodes inside the network are far more dangerous than the attacks from outside the network, and these attacks are much harder to detect because they come from the compromised nodes, whichbehave well before they are compromised. 2.4.5 Lack of Centralized Management Facility

Ad hoc networks do not have a centralized piece of managementmachinery such as a name server, which lead to some vulnerableproblems. Due to absence of centralized management facilityproblems detection of attacks, path breakages, transmission impairments and packet dropping, breakage of the cooperativealgorithm take place because decision making process isdecentralized.

Threats from Compromised 2.4.6 Restricted Power Supply nodes Inside the Network Some or all of the nodes in a MANET may Since mobile nodes are autonomous units rely on batteries orother exhaustible means that can join or leave the network with free- for their energy. For these nodes, themost imdom, it is hard for the nodes themselves to portant system design criteria for optimizawork out some eective policies to prevent tion may be energy conservation. The probthe possible malicious behaviors from all the lem that may be caused by the restricted nodes it communicate with because of the be- power supply is denial-of-service attacks . havioral diversity of dierent nodes. Further- Since the adversary knows that the target more, because of the mobility of the ad hoc node is batteryrestricted,either it can continnetwork, compromised node can frequently uously send additional packetsto the target 6

and ask it routing those additional packets, or it can induce the target to be trapped in some kind of timeconsuming computations. In this way, the battery power of the target node will be exhausted by these meaningless tasks, and thus the target node will be out of service to all the benign service requests since it has run out of power. 2.4.7 Scalability

Scalability is the problem in the mobile ad hoc network .Unlike the traditional wired network in that its scale is generallypredened when it is designed and will not change much duringthe use, the scale of the ad hoc network keeps changing allthe time: because of the mobility of the nodes in the mobile adhoc network, you can hardly predict how many nodes there willbe in the network in the future. As a result, the protocols andservices that are applied to the ad hoc network such as routingprotocol and key management servicet should be compatibleto the continuously changing scale of the ad hoc network.

layer needs. Simulations demonstrate performance Natural orman-made disasters can partition networks while threatening humanlives. Because conventional Mobile Ad-Hoc Networks MANETs cannot route messages across partitions, they may not adequately support relief eorts. To forward messages across partitions, delay-tolerant networks (DTNs) exploit in-network storage and mobility. Previous DTN routing protocols either opportunistically use, but do not modify, nodes mobility, or require dedicated mobile gateways. This paper contributes a new, cross-layer DTN routing approach based on the observation that benets of our approach in a variety of scenarios. 2.5.2 Introduction

2.5

Integration of ApplicationLayer Scheduling and Routing

Abstract

2.5.1

application-layer ordersfrom a MANETs leader also control workers mobility and ability to forward messages. Our approach attempts to minimize deadline misses and energy consumption by scheduling worker tasks considering both applicationand network7

Mobile ad-hoc networks MANETs are wireless networks whose hosts are also routers. In addition to processing their own applications, MANET nodes forward packets destined to other nodes. MANETs enable communication when network infrastructure does not exist e.g., in military tactical communication or has been damaged or compromised e.g., because of a hurricane or terrorist attack. This paper considers the problem of how to route packets in sparse MANETs with a leader. In such networks, a distinguished node, the leader, assigns tasks to and receives reports from other nodes, the workers.This work was supported in part by the SecureCITI project, funded by NSF ITR medium grant ANI-0325353. We performed this research as part of Secure CITI , an interdisciplinary project where computer and social

scientists have joined forces with the Emergency Operations Manager of the county surrounding the City of Pittsburgh to design guidelines and software elements for aiding decision-making during emergencies. Our motivating scenario is an emergency, such as a re, landslide, or ood. We assume that emergency-response teams use MANETs to communicate and each have a leader e.g., a re chief that schedules the tasks of workers e.g., reghters. In many cases, MANET nodes are sparsely distributed or move such that end-to-end routes may not always exist. However, most MANET routing algorithms fail to discover a route or drop packets sent when there is no end-to-end path between sender and receiver. Therefore, MANETs are often unsuitable for hard-real-time systems, but may be useful in soft-real-time situations, such as emergency and disaster recovery, where meeting as many deadlines as possible is a goal. 2.5.3 Network and application model

This section describes our assumptions about the network, its applications, and performance metrics. We assume that the network has a leader node and n worker nodes. The leader receives reports and responds to them by sending task assignments to workers. We consider that each task has an expected processing time and a deadline, and signicant losses result from deadline misses. For example, a beach patrol leader may receive 2.5.4 Courier scheduling algorithms report of a drowning, and respond by assigning a cardio-pulmonary resuscitation task Algorithms that we compare in our simulato a lifeguard. This task needs to be per- tions. The leader uses one of these algorithms 8

formed by a deadline, otherwise a life may be lost. We assume that, after joining a network with a leader, a worker moves or performs a task only as result of an assignment from the leader. Therefore, the leader always knows approximately where its workers are supposed to be. We assume that the leader and workers have a map of the networks area. The map is annotated with landmarks that leader and workers can identify and report to each other, so as to convey approximate position information. Alternatively, the map is annotated with geographic coordinates that nodes can match to actual GPS measurements. We further assume that assignments received from the leader may cause the network to become partitioned. The leader may send one or more workers to a site that, when reached, will lack an end-to-end route to the leader. In such a case, the network is split between the main partition, containing the leader, and a subordinate partition, containing the remote workers. Nodes within each partition communicate with each other using conventional multi-hop routing protocols. When the leader needs to communicate with a worker Ws who is in a subordinate partition, the leader sends a courier assignment to a worker Wm within the main partition. Wm physically moves between partitions so as to be able to forward task assignments to Ws and receive reports from Ws or other workers and forward them back to the leader.

when the leader needs to communicate with a node in a subordinate partition. The algorithms dier in how the leader selects the courier Wm: random courier This is the baseline algorithm. It picks Wm randomly from among the workers in the main partition. The main advantage of this algorithm is that it requires only time and does not need information about node locations, speeds, or current assignments. dedicated courier This algorithm sets aside ncnodes from the main partition for use as couriers. Similarly to dedicated mobile gateways used in previous works, such as, dedicated couriers perform no application-layer tasks, and only dedicated couriers perform courier assignments. The leader selects the dedicated courier that is closest to any worker in the destinations partition or, in case of a tie, the dedicated courier with the lowest identier. If there are O(ns) nodes in the destination partition, selection of the closest edicated courier by exhaustive search takes O(ncns) time. closest courier This algorithm picks as courier the worker Wm in the main partition that happens to be the closest to any worker in the destination partition. If there are O(nm) workers. 2.5.5 Evaluation Model

We implemented the four courier scheduling algorithms described in the previous section in the ns-2 simulator. In this section, we describe our implementation and report simulation results. We describe in this subsection our implementations data structures, messages, timers, and node behavior in response to message reception and timeouts. In the 9

description, we interchangeably use message for bundle and work order for task assignment. The leader uses a graph data structure, Graph, to keep track of network connectivity. Each worker reports its position to the leader every reportPositionInterval. The leader maintains a list of dispatched couriers dispatchedNodes, where a node is inserted when it is selected to courier a message and removed when it returns to its home position in the main partition. The leader also has a list of dedicated couriers dedicatedCouriers.A work order is queued at the leader if its assigned node has been dispatched to courier a message, or there are no reachable couriers to deliver the work order to its destination. The leader maintains a queue pendingOrdersQueue to hold these pending work orders. PendingOrdersQueue is large enough that it doesnot overow. On the other hand, each courier nodemaintains a list remoteWorkOrders of work orders that it has been assigned to deliver. Workers receive workOrder messages from the leader. Each workOrder message has a destination node workDestination as well as a task description, including deadline and processing time. Upon reception of a workOrder, a node sends an Ack message to the sender, which can be the leader or another node forwarding the message. Then, it checks if it is the workDestination. If so, it ensures that it has not received the workOrder message before and that the work order has not expired. It immediately starts executing the new task and drops the task being executed. If the node is not the destination of the workOrder message, it forwards the message toward the ultimate workDesti.

2.5.6

Conclusions

2.6

Multimedia Applications for MANETs over Homogeneous and Heterogeneou Mobile Devices
Introduction

2.6.1

Natural or man-made disasters can disrupt communications, making rescue and recovery missions moredicult. MANETs have great potential for providing connectivity in such situations, but ordinarily cannot route packets across network partitions, which are common in emergencies. Previous DTN routing protocols can overcome this limitation under certain conditions, such as favorable node mobility patterns or availability of dedicated mobile gateways.We propose a new, crosslayer DTN routing approach for MANETs with leader, such as typical emergency-response teams. In our approach, the leader schedules workers tasks considering both application- and network-layer needs. Thus, a worker may be assigned not only application-layer tasks, but also courier tasks, whose primary purpose is to provide forwarding needed for the networks operation. Our simulations show that cross-layer DTN routing can result in fewer missed deadlines and less distance traveled (and consequently, greater network uptime) than DTN routing algorithms that allocate or schedule mobile gateways in disregard of applicationlayer load.

Mobile Ad Hoc Networks are considered a vital part in beyond third generation wireless networks Nicopolitidis et al., 2003. In the matter of fact, they present a new wireless networking paradigm. Any sort of xed infrastructure is not used byMANETs. They are important sorts of WLANs, therefore, in a distributed and a cooperative environment, MANETs do eciently function . MANETs are networks of self-creating since there is a lack of routers, conguration prior to the network setup, Access Points (APs) and predetermined topology (Wu et al., 2007). MANETs are as well networks of self-administering and self-organizing. This is because in the network creation process, there is no application for central control. On MANETs, it is extremely hard to apply any of the central administration types, for instance, congestion control due to the dynamic nature of the network topology inMANETs, authentication or central routing. In short, several important applications beneted from MANETs, for example, in military, ubiquitous, emergency and collaboration computing. In this chapter, describe the necessary background for the MANETs over homogeneous and heterogeneous mobile devices. The researcher begin this chapter to introduce the related background and main concepts of the Mobile Ad Hoc Network (MANETs)

10

in Section, and explained briey about the existing wireless mobile network approaches, wireless ad hoc networks, wireless mobile approaches in Section. The characteristic of MANETs are in Section. The types of Mobile Ad hoc network in Section. The trafc types in ad hoc networks which include the Infrastructure wireless LAN and ad hoc wireless LAN are presented in Section. In Section highlight the relevant details about the ad hoc network routing protocol performance issues. The types of ad hoc protocols such as (Table-driven, On-demand and Hybrid) and Compare between Proactive versus Reactive and Clustering versus Hierarchical respectively. The existing ad hoc protocols are presented in Section . The four important issues signicant in MANET are Mobility, QoS Provisioning, Multicasting and Security is presented in Section Furthermore, the practical application and the MANET layers are shown in Section. 2.6.2 Overview of MANETs

bile nodes since the AP is the source that does communication through a single hop manner. Wireless network standards are included by the WLAN implementations and developed by Institute of Electrical and Electronics Engineers (IEEE) 802 project (IEEE 802.11, IEEE 802.11b, IEEE 802.11g, IEEE 802.11a, and IEEE 802.11n) and High Performance Radio Local Area Network Type 2 (HiperLAN2). In addition, the European Telecommunications Standardization Institute (ETSI) Broadband Radio Access Networks (BRAN) project (ETSI, 1999) developed the European version of IEEE 802.11a. A frequency of 2.4GHz runs for these standards. However, 5GHz runs for the IEEE 802.11a. For these standards, the transmission rates (bandwidths) are 2 Mbps where as for IEEE 802.11a and IEEE 802.11g, 54 Mbps is run. For IEEE 802.11b, 11 Mbps is run and for IEEE 802.11n, 100 Mbps is run. Note that a single hop WLAN with one WLAN.

2.7

The main concept of Wireless Local Area Networks (WLANs) refers to MANETs which are also called either infrastructure-based wireless networks or a single hop network Inside a WLAN, the transmission is governed by at least one xed Access Point (AP) between dierent mobile nodes. An existing network backbone and the stations contain a bridge as AP functions (Basagni et al., 2004). Both QoS and security issues are eciently controlled by the AP within a particular network. Inside the network of WLAN, there is no need for dierent mo-

Probabilistic, Application Layer Service Discovery for MANETs and Hybrid Wired-Wireless Networks
Abstract

2.7.1

Over the past years, Mobile ad-hoc networks (MANETs) have attracted a considerable degree of research attention, with service discovery, selection and invocation being among the topics of interest of previous eorts. In this paper we introduce ADDER, a probabilistic, hybrid, directory-less service discov-

11

ery mechanism. It has been designed for military IPv6-based MANETs but will work in any hybrid wiredwireless deployment. It achieves very low service acquisition time through the exchange of a very small number of short messages. Propagation of service descriptions is based on a distance vector algorithm, achieving loop and starvation freedom through a feasibility condition, which has been adopted from established and welltested routing protocols. This paper also presents evaluation results, obtained by actual execution of the ADDER daemon on two dierent test beds. The experiments aim to demonstrate that the mechanism achieves good scalability with increasing number of services and network size. 2.7.2 Introduction

In this paper we introduce ADDER, a novel, lightweight and ecient service discovery (SD) mechanism. It is the outcome of a project on military ad-hoc networks. Those networks are composed of nodes with high variance in mobility as well as power supply. For example, foot soldier carry equipment with limited power supply, while it can be assumed that ground vehicles and aircraft have an on-board energy source which can be used to power communication equipment. However, vehicles and aircraft are highly mobile and often follow unpredictable movement patterns. With that in mind, we have designed ADDER to be energy ecient by limiting the size and number of messages exchanged between nodes. in order to cope with high speeds and mobility, the mecha-

nism has low service acquisition time. Military deployments often span across multiple ad-hoc partitions (e.g. foot squads) interconnected over a wired infrastructure network. The resulting diversity in underlying technologies (routing protocols, layer 2 mechanisms) makes cross-layer approaches unsuitable. In order to cope with this diversity, ADDER has been designed as an application layer service. Lastly, ADDER adopts a loop-free, starvation free forwarding algorithm in order to propagate service information throughout the network.Additionally, in order to cope with high speeds and mobility, the mechanism has low service acquisition time. Military deployments often span across multiple ad-hoc partitions (e.g. foot squads) interconnected over a wired infrastructure network. The resulting diversity in underlying technologies (routing protocols, layer 2 mechanisms) makes cross-layer approaches unsuitable. In order to cope with this diversity, ADDER has been designed as an application layer service. Lastly, ADDER adopts a loop-free, starvation free forwarding algorithm in order to propagate service information throughout the network. 2.7.3 Conclusion

We are planning to extend our evaluation in a larger scale wireless test bed. Additionally, comparisons with other approaches are going to be conducted in a simulated environment, where we can also evaluate ADDER in very large scale deployments and validate its behaviour in mobile scenarios ADDER is currently hybrid; it always operates in reactive

12

mode, while proactive mode can be turned on optionally on all or part of the nodes. As part of our future work, we are planning to make it adaptive. Proactive mode will be turned on and o automatically at runtime, depending on node speed, mobility and energy. For example, on xed nodes with unlimited supply of energy, proactive mode will be automatically enabled. On the other hand, low power or high mobility nodes will be able to automatically switch o proactive mode without user intervention. Lastly, we are currently investigating various alternative algorithms for the adaptation of p over time on a per-service basis and evaluating their impact on ADDER sperformance.

use the Abstract MAC Layer to study the new problem of Multi-Message Broadcast, a generalization of standard single-message broadcast, in which any number of messages arrive at any processes at any times. We present and analyze two algorithms for MultiMessage Broadcast in static networks: a simple greedy algorithm and one that uses regional leaders. We then indicate how these results can be extended to mobile networks.

3.1

Selective Bit-error Checking at the MAC Layer for Voice Over Mobile Ad Hoc Networks
Abstract

Mac Layer In Manet

3.1.1

A diversity of possible communication assumptions complicates the study of algorithms and lower bounds for radio networks. We address this problem by dening an Abstract MAC Layer. This service provides reliable local broadcast communication, with timing guarantees stated in terms of a collection of abstract delay functions applied to the relevant contention. Algorithm designers can analyze their algorithms in terms of these functions, independently of specic channel behavior. Concrete implementations of the Abstract MAC Layer over basic radio network models generate concrete denitions for these delay functions, automatically adapting bounds proven for the abstract service to bounds for the specic radio network under consideration. To illustrate this approach, we

Mobile ad hoc networks (MANET) have more severe operating conditions than traditional wireless networks. The MAC protocol of IEEE 802.11 mitigates collisions and ensures error-free packet transmissions at the cost of limiting capacity and increasing latency. For voice transmission over MANETs this cost should be minimized.We propose and examine selective error checking (SEC) at the MAC layer of 802.11 that takes advantage of the fact that many of the speech bits can tolerate errors while other bits must be protected for eective reconstruction of the speech. Simulation results demonstrate that the network performance and the speech quality are substantially improved by modifying the MAC layer with SEC to suit a particular GSM speech compression standard, the Narrow-Band Adaptive MultiRate (NB-

13

AMR) coder operating at a rate of 7.95 kbps. 3.1.3

Voice Transmission Over Ad Hoc Networks

3.1.2

Introduction

A mobile ad hoc network (MANET) is a wireless LAN (WLAN) wherein mobile nodes can communicate with one other without relying on any pre-existing infrastructure. In a MANET, all the mobile nodes have equal capabilities and operate not only as hosts but also as network routers. Communication links might be broken and a dynamic routing protocol is needed. Dynamic routing requires a nontrivial amount of overhead trac on the network to discover changes in the available paths or to identify on demand a specic route when needed. The eciency of the routing protocol and other higher layer protocols such as IP can indirectly depend on the underlying medium access control (MAC) protocol. The IEEE 802.11 [1], includes a MAC protocol and a physical (PHY) layer protocol. The MAC protocol mitigates collisions and ensures error-free packet transmissions at the cost of limiting capacity and increasing latency. Many research eorts have concentrated on improving the performance and capacity of the 802.11 MAC protocol for data. Also, some studies have focused on modifying the IEEE 802.11 MAC protocol to support real-time trac, such as voice, in MANETs. In general, the performance of real-time communication in MANETs can be improved by reserving bandwidth.

A simple MANET with three colinear nodes is shown in g. Nodes A and C are out of range for either direction of transmission (e.g., they cannot hear each other, while A and B and also B and C can hear each other). For voice communication between nodes A and C, an intermediate nodeIII. VOICE TRANSMISSION OVER AD HOC NETWORKS A simple MANET with three colinear nodes is shown in Fig. 1. Nodes A and C are out of range for either direction of transmission (e.g., they cannot hear each other, while A and B and also B and C can hear each other). For voice communication between nodes A and C, an intermediate node is needed to forward the packet and B serves this role here. Let A be the source node where the speech signal is generated, compressed and packetized with the UDP/IP protocol stack. The MAC frame in this case consists of the voice data, the UDP header, the IP header and the MAC header as shown in We assume for simplicity here that the nodes have a xed location for the duration of a voice call, so they are not currently mobile. However, since the nodes are capable of mobility, a routing protocol, an essential part of any MANET, operates at node A to discover node B as the desired nextnode. Each successive packet is transmitted over the wireless medium by the 802.11 physical layer. At node B, the routing protocol and the 802.11 protocol are involved to identify C as the next-node (and nal destination) and deliver the packet to C. At the destination

14

node C, the received packets are buered, reordered as needed, unpacketized, and the voice data is uncompressed and speech is reconstructed. 3.1.4 Simulation and Results

A network scenario shown in Fig. 2 is examined by NS- 2 simulator [15] in this research. The distances between thenodes are chosen so that the nodes are at a hearing distancefrom each other and share the medium. The data rate of IEEE 802.11 was congured to be 2 Mbps. In our simulation model, the nodes have no mobility. This is primarily because our interest in this paper is to focus on the eectiveness of SEC at the MAC layer. Hence, we assume that the network is xed for the duration of the voice transmission, yet the normal overhead trac due to dynamic routing protocol is present. Specically, we use the destination sequenced distance vector (DSDV) routing protocol. To evaluate the eectiveness of the new MAC protocol for multihop transmission, one voice transmission is separately simulated from node 1 to A, 1 to B, 1 to C, and 1 to 2. To evaluate the eectiveness of the new MAC protocol for multiple trac transmission, 1, 2, 3, and 4 voice transmissions from node 1 to node B are separately simulated. The Elliott-Gilbert two state Markov model [17], [18], shown in g, is used to model the wireless channel. The bit errors generated by this model are introduced to MAC frames. In this model, each state represents a binary symmetric channel. Bit errors occur with low probability pG in the good state (G), and bit errors happen with

high probability pB in the bad state (B). pGB and pBG represent the probability of switching from the good state to the bad state and vice versa. An average link proposed in [19] is used in our simulation where pG = 0, pB = 0.33, pGB = 0.01 and pBG = 0.14. Based on the Elliott-Gilbert model, this results in an average error probability for this link of 0.022. Voice TransmissionFor our simulation, we use 25 seconds of speech, consisting of 4 male sentences and 4 female sentences modeled as a constant bit rate trac source with 20 ms of speech for each packet. After the speech is compressed by the NB-AMR coder at 7.95 kbps, there are 159 bits (20 bytes) for a 20 ms.

3.1.5

Conclusions

In this paper, we proposed the SEC mechanism at the IEEE 802.11 MAC layer for voice over MANETs. We have shown that SEC improves the network performance and enhances the speech quality. While further research is needed to make voice over MANETs practical, it is evident that SEC can be combined with many other schemes for voice over MANETs, such as reservation schemes or high priority schemes, header compression, error concealment and silence compression. SEC is simple to implement and it is eective in conserving bandwidth and reducing latency for voice communication over manets.

15

3.2

AN ENHANCED MAC been widely used. A portable device usuPROTOCOL STACK FOR ally has several wireless interfaces, such as IEEE 802.11 Wireless LAN (WLAN), GenMANETs
ABSTRACT

3.2.1

This paper deals with design goals to provide seamless connection to the Internet when mobile hosts roam between MANETs and to achieve load-balancing routing when mobile hosts have multiple gateways available. We address three issues in this paper namely IP address resolution, mobility management under NAT structure, and load-balancing routing. Mobile IP and NAT traversal are adopted to provide seamless roaming capability in private networks. A load-balancing routing protocol was implemented to relieve the bottleneck problem. A prototype is implemented on Windows platform to verify our architecture and test results do show the benet of load-balancing routing. INTRODUCTION Due to its exibility, Mobile Ad Hoc Network (MANET) has attracted a lot of attention recently. Most existing works, however, limit a MANET as a stand-alone network. In this paper, we propose a multi-tier MANET by extending the connectivity of the MANET to the Internet and voice-networks. Some hosts in the MANET are equipped with cellular interfaces and are called gateways, which can provide Internet connections. Such extension would greatly improve the connectivity of MANET. With the advance of embedded computing technologies, portable devices, such as laptops, Personal Digital Assistants (PDAs), and cellular phones, have

eral Packet Radio Service (GPRS), Personal Handy-phone System (PHS), and/or Bluetooth. Wireless communications are typically supported in two models: infrastructure and ad hoc. Of these two options, forming a mobile ad hoc network(MANET) is more exible since it is independent of the availability of base stations. Hence, intensiveresearch has been dedicated to MANET [1, 2]. A MANET is typically considered as a stand-alone network. However, it is important to enable its Internet accessibility. On one hand, users in a MANET can enjoy the tremendous resources in the Internet. On the other hand, the connectivity between multiple MANETs may be greatly improved. For such connectivity, several works [4, 5, 6, 7] have proposed possible architectures by deploying gateways to help mobile hosts route packets to the Internet. Among these approaches, some takes a proactive approach by modifying DSDV , some takes a reactive approach by modifying AODV, while some takes a hybrid approach. In this paper, we propose a multitier MANET architecture in which broadband WLANs (such as IEEE 802.11 a/b/g) are equipped in all mobile stations to form the low-tier network, and cellular interfaces (such as GPRS/PHS/3G) are equipped in some stations to form the high-tier network. Stations with high-tier interfaces are called gateways and can connect to the Internet. Depending on its service range, each gateway together with the stations whose Internet connections are supported by the gateway

16

constitutes a sub-MANET. The MANET under our scope is a collection of multiple subMANETs. In this paper, we address several important design issues in MANET architecture. 3.2.2

while roaming. Note that the high-tier interfaces may also use private IP addresses, making traditional Mobile IP unusable. Several solutions [9, 10, 11] have been proposed to support IP mobility under private networks. We apply the NAT traversal mechanism [9] to NETWORK ARCHITECTURE achieveseamless roaming in private networks. 3.2.3 PROTOCOL DESIGN

We consider a set of mobile hosts forming a MANETs. Each host is equipped with an IEEE 802.11 WLANcard, and these interfaces form the low-tier network. The Destination-Sequenced Distance Vector (DSDV) routing protocol [3], which adopts a proactive approach, is used on the low-tier network. A number of hosts are designated as gateways. Each gateway host is equipped with an extra cellular interface, such as PHS or GPRS, which enables the host to access infrastructure networks (and thus the Internet). Cellular interfaces with Internet access capability form the high-tier network. Note that these interfaces can be heterogeneous. The MANET can be physically connected (through the low- or high-tier network) or disconnected. Each gateway together with the stations whose Internet connections are supported by it is called a sub-MANET. A set of stations that forms a connected component but does not have a gateway inside is disconnected from the MANET and is not allowed to connect to the Internet. When the network topology changes or then gateways change their points of attachment, hando procedure may be taken. To support seamless roaming, we adopt Mobile IP with the colocated address mode [8]. Hosts rely on their home agents to maintain their connections

Our design goal is to provide seamless connection to the Internet when mobile hosts roam between MANETs and to achieve loadbalancing routing when mobile hosts have multiple gateways available. Weaddress three issues: IP address resolution, mobility management under NAT structure, and loadbalancing routing. However, the full treatment of each of these issues is beyond the scope of this paper. A DHCP server is installed in each gateway. Before communicating with other hosts, a mobile host needs to retrieve an IP address from a gateway. To avoid confusion, we assign an exclusive section of IP addresses to each DHCP server. Note that we also allow a host to use its old IP address after roaming into a new sub-MANET. When a new mobile host n joins the MANET, it rst broadcasts a DHCPd iscover. 3.2.4 CONCLUSION

In the literature, most works consider a MANET as a stand-alone network. In this paper, we design a multitier MANET, considering gateways as bridges to the Internet. This greatly improves the connectivity

17

of MANET because cellular networks nowadays are almost globally available. However, Internet connectivity does cause several problems: address conguration, connection maintenance when roaming, and trafc bottleneck on gateways. In this paper, we show how to congure a MANET as a private network and modify DHCP to avoid possible broadcast storms. Mobile IP and NAT traversal are adopted to provide seamless roaming capability in private networks. A load-balancing routing protocol was implemented to relieve the bottleneck problem. A prototype is implemented on Windows platform to verify our architecture and testing results do show the benet of load-balancing routing

3.3

A novel ecient powersaving MAC protocol for multi-hop MANETs

INTRODUCTION

3.3.1

Following recent improvements in the performance of wireless communication systems, mobile ad hoc networks (MANETs) [1] have become increasingly important in increasingly wide range of applications, such as battleelds and other military environments, disaster areas, and outdoor activities. AMANET is a multi-hop wireless network that is formed dynamically from an accumulation of mobile nodes without the assistance of a centralized coordinator. As the radio propagation range is limited, each mobile node has only limited information, such as its own ID and the Medium Access Con-

trol (MAC) address of its one-hop neighbors. Therefore, if two nodes are not within the radio propagation range, a multi-hop, via one or more intermediate nodes, is required to forward packets. The forward function of each intermediate node consumes time and resources, such as power and bandwidth. However, a mobile node has limited power. This study addresses the maximization of the lifetime of mobile nodes through various mechanisms. The power consumption of a battery in a mobile node must be minimized to maximize its lifetime [2, 3]; otherwise, the battery may quickly run out of power, making the mobile node useless. The operating states of a network interface can be categorized into transmit, receive, idle, and sleep states, and the estimated power consumption of each state is as presented in Table I. An interface in the sleep state can neither transmit nor receive any packets, and thus this state consumes the lowest power. To be able to transmit and receive packets, an interface must be woken up. A mobile node that is awake, but neither transmitting nor receiving data, is said to be idle. A node consumes the most power when it is in the awake state. Therefore, the proposed power-ecient protocol depends on mobile nodes staying in the sleep state most of the time, unless data have to be transmitted. The reduction of power consumption by MANETs has been studied widely. Existing powersaving MAC protocols can be classied into two categoriessynchronous wake up approaches [48] and asynchronous wake up approaches [1015]. In synchronous wake up approaches, all nodes must execute a clock synchronization mechanism

18

[4, 1620]. Asynchronous wake up approaches require no such synchronization mechanism. However, the neighbor discovery time is the most important issue in asynchronous wake up approaches. They must adjust the overlap of a nodes wake up time with that of its neighbors, resulting in increased power consumption and long transmission delay. Thus, this study focuses on the synchronous wake up approach. 3.3.2 PRELIMINARIES

Various power-saving protocols for IEEE 802.11 wireless local area network have recently been proposed. This section briey reviews several power-saving protocols [415] and discusses some of the problems associated with MANETs, as the synchronous power-saving approaches, which require an eective time synchronization mechanism, is considered. Section 2.2 also reviews numerous time synchronization mechanisms [4, 1620].reviews of power-saving protocols Synchronous wake up approaches. The most well-known synchronous wake up powersaving protocol is the IEEE 802.11 standard [4], which was originally designed for single-hop ad hoc networks. As shown in Figure 1, time is divided into beacon intervals. In the PSM of the IEEE 802.11 standard, all nodes are synchronized by transmitting beacon frames to one-hop neighbors at the beginning of the beacon interval. After the beacon frame has been sent, the node sends an ad hoc Trafc Indication Map (ATIM) frame to inform other nodes that it has packets that are waiting to be transmitted during the ATIM win-

dow. Upon receiving an ATIM-ACK frame from the destination node, a node obtains the right of transmission and begins to transmit data immediately after the ATIM window ends. Both sender and destination nodes are awake during the transmission period. Otherwise, at the end of ATIM window, a node enters the power-saving state. IEEE 802.11 PSM has been extended to multi-hop MANETs [5], to activate paths, minimize delay, and conserve energy. However, the proposed synchronization strategy, routing strategy, and power management capability depend on extra support from MAC layer. Additionally, the potential problem of network partitioning has not been addressed. Span [6] is based on the notion of a dominating set and extends the sleep time of mobile hosts to reduce power consumption. Span adaptively elects coordinators to generate a connected domination set; they are kept awake at all times to perform low-latency multi-hop routing. Other noncoordinators go through periodic cycles of sleep and wakefulness and periodically check whether they should wake up and become coordinators. Although Span guarantees ecient energy consumption and low delay latency in dense networks, it has two limitations. One is that coordinators must remain active at all times, broadcasting HELLO messages to maintain the backbone, increasing the overhead. The other is its synchronization overhead. Special-purpose methods for reducing power consumption of MANETs have been proposed [7, 8]. A node can power down during its natural silent periods [7]: when a node does not expect to transmit, receive, or relay packets, it can

19

power o its network interface. Trac aware PSM (TA-PSM) [8] also achieves good performance with a light trac load. TA-PSM allows the node directly to enter the doze state when it does not need to transmit or receive packets, even if a beacon or ATIM frame has to be sent. Instead of entering the idle state of IEEE 802.11 PSM, the node enters a doze state to save more power. However, such approaches depend on the monitoring of trac at each node to guarantee transmission throughput and low transmission latency. Hence, these approaches may be not suitable for heavy trac scenarios.

3.4

MAC Improvements for MANET

3.4.1

ABSTRACT

3.3.3

CONCLUSIONS TURE WORK

AND

FU-

Power conservation is very important to prolong the battery life of important devices. This work proposed a novel ecient power-saving MAC protocol for multihop MANETs, called p-MANET. p-MANET consists of three mechanismsthe hibernation mechanism to prevent the consumption of power for unnecessary tasks, the beacon inhibition mechanism solves beacon storm problem, and the low-latency next hop selection mechanism oers heuristic strategies to select eciently the next-hop node for packet forwarding. To conrm the eectiveness of pMANET, we present the theoretical analysis related to p-MANET concerning the average awake time and average delay time. And extensive simulations were performed, and the results revealed a power saving of over 70

Broadcasting is one of the essential communication models of MANETs. Many MANET multicast routing protocols rely heavily upon MAC layers broadcast support. However, the broadcast mechanism of the standard IEEE 802.11 cannot provide reliable broadcasting service. In this paper, we improve the IEEE 802.11 broadcast mechanisms reliability by introducing the new layer of MAC called Dual MAC. Multihop ad-hoc wireless networks oer great challenges for protocol designers. Stations in such networks are constrained by factors like low power, limited bandwidth, link errors, and collisions. Changes are needed at various levels of the protocol stack, most importantly at the medium access layer (MAC). The medium access mechanism in multihop wireless networks should minimize collisions, and take care of the hidden and exposed node problems. The IEEE 802.11 MAC with Distributed Coordination Function (DCF) does not scale well in such networks. We introduce Point Coordination Function (PCF) in the region of high trac areas, and discuss its eect on network performance. To improve network scalability and throughput, we propose the design of a new MAC called Dual MAC. This work discusses architecture and working of the dual MAC in detail.

20

3.4.2

INTRODUCTION

tion) is required.

In recent times, the wireless networks have become very popular.Wireless LANs are being deployed on airports, conferences, etc. People have started using portable laptops to access Internet and other resources using wireless networks while moving. Another area which has generated a lot of interest recently, is wireless adhoc networks. An adhoc network is formed when two or more stations come together form an independent network. Ad-hoc networks are also termed as infrastructure-less networks since as they do not require any prior infrastructure. Two stations that are within transmission range of each other are called one hop neighbors. Multihop ad-hoc networks are ones in which the stations can talk to stations more than one hop away via intermediate stations. Cooperative ad-hoc networks are formed by several homogeneous wireless stations. All the stations cooperate with each other, i.e., the trafc for the stations that are more than one hop away is routed by the intermediate stations. The intermediate stations are called relaying stations If the channel is still idle at the end of the CONTENTION period the node transmits its packet otherwise it repeats the process dened in 3 above until it gets a free channel. Key: D = DCF Inter Frame Space (DIFS) S = Short Inter Frame Space (SIFS) CW = Contention Window MPDU = MAC Protocol Data Unit A = Ack 802 11 also offers a polling mode (known as PCF - Point Coordination Function) which is fairly classic polling scheme e.g. 3270 bi-sync!! As with all polling protocols a single master (Base Sta-

3.4.3

ORIGINAL MAC

The basic 802.11 MAC layer uses the Distributed Coordination Function (DCF) to share the medium between multiple stations. DCF relies on CSMA/CA and optional 802.11 RTS/CTS to share the medium between stations. This has several limitations: If many stations communicate at the same time, many collisions will occur, which will lower the available bandwidth (just like in Ethernet, which uses CSMA/CD)The original 802.11 MAC denes another coordination function called the Point Coordination Function (PCF): this is available only in infrastructure mode, where stations are connected to the network through an Access Point (AP). This mode is optional, and only very few APs or Wi-Fi adapters actually implement it. APs send beacon frames at regular intervals (usually every 0.1 second). Between these beacon frames, PCF denes two periods: the Contention Free Period (CFP) and the Contention Period (CP). In CP, the DCF is simply used. In CFP, the AP sends Contention Free-Poll (CF-Poll) packets to each station, one at a time, to give them the right to send a packet. The AP is the coordinator. This allows for a better management of the QoS. Unfortunately, the PCF has limited support and a number of limitations (for example, it does not dene classes of trac).

21

3.4.4

MAC IN IN MULTIHOP SCE- 802.11 MAC protocol using the PCF and NARIO DCF mechanisms. We nd that without modications, the PCF and DCF are not very The IEEE 802.11 MAC is designed for wire- useful in multihop networks. The dual MAC less LANs. The requirements of multihop ad- was designed to eliminate exposed and hidhoc networks are more challenging than those den node problems in the central region of a of wireless LANs. We will investigate the op- centralized multihop network. eration of IEEE 802.11 MAC in centralized multihop ad-hoc networks. The terms station and node are used interchangeably throughout the thesis. Multihop cooperative wire- 4 Network Layer less ad-hoc networks will be simply referred to as multihopnetworks. Consider a multi- At the network layer, the routing protocols hop centralized scenario, as shown in the g- delay in detecting topology changes may lead ure 4.1. For convenience, the stations inside to periods without connectivity and a risk of the network are classied into following cate- loops, both in case of mobility and uctuatgories: Central station is the central control- ing links. Also, the end-to-end transmission ling station. Most of the trac in the net- time/RTT10 will change as a result of changwork is directed towards it. Inner stations ing paths between the source and destinaare within one hop boundary of the central tion. If the RTT is increased too much, timestation. Boundary stations are at one hop outs will occur on the TCP sender, causing boundary of the central station. These sta- unnecessary retransmissions. If two neightions act as relaying stations for the stations boring nodes have dierent relative mobiloutside the reach of central node. Outer sta- ity, they will eventually become disconnected. tions are outside the communication range of Any routes using this link will fail, and it central node. is the task of the routing protocol to de3.4.5 CONCLUSION tect the link break and discover an alternative route between the source and destination. In a MANET, this kind of topology change will happen on a fairly frequent basis, due to the limited communication range of radios. Route failures and route changes may impact TCP in several ways. Route failures can cause packet drops at the intermediate nodes. These will be interpreted as congestion loss, a timeout event happens and TCP enters the slow-start process as if congestion occurred. Even if the routing protocol is able

The design of a MAC that meets the demand of a multihop wireless network is great challenge. The restrictions like limited bandwidth, low power, and limited transmission range make this challenge even greater. Further, the hidden and exposed node problem oer even more diculties by increasing the chance of collision. In this work, we have investigated the usefulness of IEEE

22

4.1

both analysis and simulation results, the efMobile ad hoc network (MANET) is a type fectiveness of proposed model in a MANET of wireless ad hoc network, and is a self- environment is demonstrated. conguring network of mobile devices connected by any number of wireless links. Every 4.1.2 INTRODUCTION device in a MANET is also a router because it is required to forward trac unrelated to Mobile ad hoc network (MANET) is a type its own use. Each MANET device is free to of wireless ad hoc network, and is a selfmove independently, in any arbitrary direc- conguring network of mobile devices contion, and thus each device will potentially nected by any number of wireless links. Every change its links to other devices on a regu- device in a MANET is also a router because lar basis. Such networks extend the limited it is required to forward trac unrelated to wireless transmission range of each node by- its own use. Each MANET device is free multihop packet forwarding. Security is one to move independently, in any arbitrary dicrucial requirement for these mission-critical rection, and thus each device will potentially applications. In particular, in MANET, any change its links to other devices on a regular node may compromise the routing protocol basis. Such networks extend the limited wirefunctionality by disrupting the route discov- less transmission range of each node bymultihop packet forwarding. Security is one crucial ery process. requirement for these mission-critical applications. In particular, in MANET, any node 4.1.1 Abstract may compromise the routing protocol funcThe important security issue in mobile ad hoc tionality by disrupting the route discovery networks is to protect the routing layer from process. In this paper, an important security malicious attacks. A unied security solution issue is tackled in ad hoc networks, namely for such networks is applied to protect both the protection of their network-layer operarouting and data forwarding operations in the tions from malicious attacks. Without aprouting layer. In this paper the proposed propriate protection, the malicious nodes can model does not apply he cryptographic prim- readily function as routers and prevent the itives on the routing messages This model network from correctly delivering the packprotects the network by detecting and iso- ets. For example, the malicious nodes can lating the malicious nodes. In this proposed announce incorrect routing updates which are model, every node is monitoring other near- then propagated in the network, or drop all est neighboring nodes. A novel recognition the packets passing through them. A prostrategy is applied to decrease its overhead posed model is used to protect both routing 23

A Novel security frame- as time evolves. In the proposed model inforwork for protecting Net- mation cross-validation is used to protect the network in a self-organized manner. Through work Layer operations

and packet forwarding together. The research directions towards security in MANETs are still at their infancy. Security issues arise in many dierent areas including physical security, key management, routing and intrusion detection, many of which are vital to a functional MANET. Due to their particular architecture, ad-hoc networks are more easily attacked than wired network. There are two kinds of attacks: the passive attacks and the active attacks. A passive attack does not disrupt the operation of the protocol, but tries to discover valuable information by listening to trac. Instead, an active attack injects arbitrary packets and tries to disrupt the operation of the protocol in order to limit availability, gain authentication, or attract packets destined to other nodes. The routing protocols in MANET are quite insecure because attackers can easily obtain information about network topology. Indeed in AODV and DSR protocols, the route discovery packets are carried in clear text. So a malicious node can discover the network structure just by analyzing this kind of packets and may be able to determine the role of each node in the network. With all these information more serious attacks can be performed in order to disturb the network operation by isolate important nodes, etc. The attacks in modication and impersonation are: One of the simplest ways for a malicious node to disturb the good operation of an ad-hoc network is to announce better routes (to reach other nodes or just a specic one) than the other nodes.

4.1.3

SIMULATION RESULTS AND ANALYSIS

In this section, the performance of proposed model is evaluated through extensive simulations. The simulation methodology is started and performance metrics is evaluated. The results show that proposed model is eective in protecting the network layer of ad hoc networks even in a highly mobile and hostile environment. The proposed model is implemented in the ns-2 simulator. Performance evaluations are based on the simulations of 100 wireless nodes that form an ad hoc network over a rectangular (3000 m 600 m) at space in 1500 s of simulation time. The physical layer at each networking interface is chosen to approximate the Lucent Wave LAN wireless card. The MAC layer protocol and the routing protocol are 802.11 DCF and modied AODV protocol, respectively. An improved version of random waypoint model, which is recently proposed as the mobility model. Set the minimum speed for each node as 2 m/s except for the static network case, and vary the maximum speed to evaluate the impact of node mobility on proposed model performance. The pause time is set to 0 to simulate an ad hoc network in which nodes are constantly roaming. Before the simulation runs, randomly select a certain fraction, ranging from 0 4.1.4 CONCLUSION

One fundamental challenge for security design in mobile ad hoc networks is the absence of any preexisting infrastructure support. 24

This work explores a novel self-organized approach to securing such networks. To this end, we have presented a proposed model, a network-layer security solution that protects routing and forwarding operations in a unied framework. This model exploits localized collaboration to detect and react to security threats. All nodes in a local neighborhood collaboratively monitor each other and sustain each other, and no single node is superior to the others. The proposed design is self-organized, distributed, and fully localized. Both analysis and simulations results have conrmed the eectiveness and efciency of the proposed framework in protecting the network layer in mobile ad hoc networks.

such attacks in MANETS is presented The paper also focuses on dierent security aspects of network layer and discusses the effect of the attacks in detail through a survey of approaches used for security purpose. 4.2.2 INTRODUCTION

4.2
4.2.1

Network Layer Attacks and Defense Mechanisms

ABSTRACT

The foremost concerned security issue in mobile ad hoc networks is to protect the network layer from malicious attacks, thereby identifying and preventing malicious nodes. A unied security solution is in very much need for such networks to protect both route and data forwarding operations in the network layer. Without any appropriate security solution, the malicious nodes in the network can readily act to function as routers. This will solely disturb the network operation from correct delivering of the packets, like the malicious nodes can give stale routing updates or drop all the packets passing through them. In this paper a study that will through light on

Mobile ad hoc network (MANET) is a type of wireless ad hoc network, and is a selfconguring network of mobile devices connected by any number of wireless links. Each device in a MANET is free to move independently in any direction, and will therefore change its links to other devices frequently. Each must forward trac unrelated to its own use, and therefore be a router. The primary challenge in building a MANET is equipping each device to continuously maintain the information required to properly route trac. Such networks may operate by themselves or may be connected to the larger Internet. Many academic papers evaluate protocols and abilities assuming varying degrees of mobility within a bounded space, usually with all nodes within a few hops of each other and usually with nodes sending data at a constant rate, packet drop rate, the overhead introduced by the routing protocol, and other measures. Security is an essential service for wireless network communications. However, the characteristics of MANETS pose both challenges and opportunities in achieving security goals, such as condentiality, authentication, integrity, availability, access control, and nonrepudiation [1]. The countermeasures can be considered as features or functions that reduce

25

or eliminate security vulnerabilities and attacks. First, in this paper an overview of network layer attacks is given, and then the security counter measures. Since in MANETS the nodes dynamically set up paths among themselves to transmit the packets, it is referred as infrastructure less network. The nodes in MANET can communicate directly if they are in within each others wireless transmission ranges otherwise they have to rely on some other nodes to transmit messages if the nodes are outside each others transmission range [2]. Thus, several intermediate hosts relay the packets which are sent by the source host before they reach the destination host, which in turn leads to a multi-hop scenario I.e. each node, will act as a router. The nodes cooperation is very much important for a successful communication. Thus, a MANET has several salient characteristics [3]: dynamic topologies, resource constraints, limited physical security, and no infrastructure. Possible applications of MANET include: Soldiers relaying information for situational awareness on the battleeld, business associates sharing information during a meeting; attendees using laptop computers to participate in an interactive conference; and emergency disaster relief personnel coordinating eorts after a re, hurricane, or earthquake [1]. The other possible applications [2] include personal area and home networking, location-based services, and sensor networks. There are a wide variety of attacks that target the weakness of MANETS. For example, routing messages are an important component of mobile network communications, as each packet needs to be passed

quickly through intermediate nodes, which the packet must traverse from a source to the destination. Malicious routing attacks can target the routing discovery or maintenance phase by not following the specications of the routing protocols. There are also attacks that target some particular routing protocols, such as DSR, or AODV [4] [5]. More sophisticated and subtle routing attacks have been identied in recent published papers, such as the black hole (or sinkhole) [6], Byzantine [7], and wormhole [8] [9] attacks. Currently routing security is one of the hottest research areas in MANET, so only the research initiative is taken for a specic layer like network layer in OSI model [1]. This paper is organized as follows. In Section 2, description about the network layer attacks is given. In Section 3, proposed solutions for the dierent network layer attacks are discussed, including multilayer attacks. In section 4, a discussion on open challenges and future directions is given. 4.2.3 NETWORK SECURITY ATTACKS

The connectivity of mobile nodes over a wireless link in MANETS which is multihop in nature strongly relies on the fact that ensures cooperation among the nodes in the network. Since network layer protocols forms connectivity from one hop neighbors to all other nodes in MANET, the assurance of cooperation among nodes is required. Recently variety of network layer targeted attacks have been identied and heavily studied in research papers. As a consequence of attacking network layer routing protocols, adversaries

26

can easily disturb and absorb network trac, inject themselves into the selected data The packets in the network trac could be forwarded to a suboptimal path or to a not existing path, which introduces signicant delay and packet losses in the network. The adversaries send some ctitious routing updates to create routing loops or to introduce severe congestion in some portions of the network or to make some parts of the network inaccessible. The main eect of the presence of malicious nodes in the network is excessive network control trac which intensies the network congestion and as a result the performance of the network degrades. The principle idea behind this paper is to evaluate what security measures have been considered till date for identication of malicious nodes and preventing them in the network. Through a relative study, it can be revealed the research work carried using dierent cryptographic techniques considered for the security purposes to avoid malicious nodes in MANETS. Finally it can be concluded with a note that what precautions can be applied to ensure condentiality and integrity in the network to upgrade the network performance.

layer diers such as not forwarding the packets or adding and modifying some parameters of routing messages; such as sequence number and hop count. The most basic attack executed by the nodes in the network layer is that an adversary can stop forwarding the data packets. The consequence caused by this is that, whenever the adversary is selected as an intermediate node in the selected route, it denies the communication to take place. Most of the times the black hole attack is launched by the adversaries, whenever AODV is used as the data forwarding protocol. Consider a malicious node which keeps waiting for its neighbors to initiate a RREQ packet. node receives the RREQ packet, it will immediately send a false RREP packet with a modied higher sequence number. So, that the source node assumes that node is having the fresh route towards the destination. The source node ignores the RREP packet received from other nodes and begins to send the data packets over malicious node. A malicious node takes all the routes towards itself. It does not allow forwarding any packet anywhere. This attack is called a black hole as it swallows all the data packets [13] [14]. Gray hole Attack: A variation of 4.2.4 Network Layer Attacks Descrip- black hole attack s is the gray hole attack, in which nodes either drop packets selectively tion (e.g. dropping all UDP packets while forBlack hole Attack: In routing mechanism of warding TCP packets) or drop packets in a ad hoc networks three layers namely physi- statistical manner (e.g. dropping 50 cal, MAC and network layers plays a major role. As MANETs are more vulnerable to various attacks, all these three layers suer from such attacks and cause routing disorders. The variety of attacks in the network 27

4.2.5

DEFENSE AGAINST NET- resource consumption attacks. It uses oneWORKLAYER ATTACKS way hash function without the usage of asymmetric cryptographic mechanism. The mechThe previous section reveals the possibility anism uses authentication to dierentiate beof various attacks on the network layer and tween malicious and non-malicious nodes, now the focus is on the several security mea- which in turn reduces resource consumption sures taken to overcome these attacks. As attacks launched by malicious nodes. SEAD it is a known fact, cryptography is one of avoids routing loops, but the drawback lies the most common and reliable means to en- whenever the attacker uses the same metric sure security in MANETS. The main no- and sequence number used for authentication tions for cryptography are condentiality, in- were same by the recent update message and tegrity, authentication and non-repudiation. updates with new update message. The reThe cryptography is discussed in detailed search update message from this mechanism in. MANETS have certain challenges in key is that it can also be used for other distance management due to lack of infrastructure, ab- vector routing protocols. sence of dedicated routers and mobility of nodes, limited processing power and limita4.3 A Robust Approach to Detion of battery power, bandwidth and memtect and Prevent Network ory. The main requirement to ensure security Layer Attacks in MANETS in MANETS is to have a secure routing protocol which should have properties to detect 4.3.1 Abstract malicious nodes, guarantee of exact route discovery process, maintaining condential net- A dynamic wireless network that is formed work topological information and to be self- without any pre-existing infrastructure, in stable against attacks. SAR (Secure-Aware which every node can act as a router is called Ad Hoc Routing protocol), which denes a a mobile ad hoc network (MANET). Since level of trust as a metric for routing and as MANETS has not got clear cut security proan attribute for security for routing. SAR visions, it is accessible to any of the auusing AODV uses encryption and decryption thorized network users and malicious attackprocess using a common key. The main draw- ers. The greatest challenge for the MANETS back with SAR protocol is whenever the lev- is to come with a robust security solution els of security rise; it needs dierent keys for even in the presence of malicious nodes, so dierent levels, thereby increasing the num- that MANET can be protected from varber of keys . SEAD (Secure Ecient Ad Hoc ious routing attacks. Several countermeaDistance Vector Routing protocol) is mainly sures have been proposed for theserouting designed for DSDV (Destination-Sequenced attacks in MANETS using various cryptoDistance Vector). This protocol can over- graphic techniques. But most of these mechcome DoS, all types of routing attacks and anisms are not considerably suitable for the 28

resource constraints, i.e., bandwidth limitation and battery power, since they results in heavy trac load for exchanging and verication of keys. In this paper, a new semantic security solution is provided, which suits for the dierent MANET constraints and also is robust in nature, since it is able to identify and prevent four routing attacks parallelly. The experimental analysis shows the identication and prevention of the four attacks parallelly I.e., packet dropping, message tampering, black hole attack and gray hole attack.

4.3.3

A Distributed Security Scheme

4.3.2

INTRODUCTION

A MANET has got some of the important properties like self organized and rapid deployable capability; which makes it widely used in various applications like emergency operations, battleeld communications, relief scenarios, law enforcement, public meeting, virtual class rooms and other securitysensitive computing environments. There are several issues in MANETS which addresses the areas such as IP addressing, radio interference, routing protocols, power Constraints, security, mobility management, bandwidth constraints, QOS, etc;. As of now some hot issues in MANETS can be related to the routing protocols, routing attacks, power and bandwidth constraints, and security, which have raised lot of interest in researchers. Even though in this paper we only focus on the routing attacks and security issue in MANETS.

Reliable network connectivity in wireless networks is achieved if some counter measures are taken to avoid data packet forwarding against malicious attacks. A lot of research has taken place to avoid malicious attackers like, a Survey on MANET Intrusion Detection, Advanced Detection of Selsh or Malicious Nodes in Ad hoc Networks , Detecting Network Intrusions via Sampling : A Game Theoretic Approach , Collaborative security architecture for black hole attack prevention in mobile ad hoc networks , A Distributed Security Scheme for Ad Hoc Networks [6], Wormhole attacks detection in wireless ad hoc networks: a statistical analysis approach [12], Enhanced Intrusion Detection System for Discovering Malicious Nodes in Mobile Ad Hoc Networks Detection and Accusation of Packet Forwarding Misbehavior in Mobile Ad- Hoc networks WAP: Wormhole Attack Prevention Algorithm in Mobile Ad Hoc Networks A Reliable and Secure Fra1mework for Detection and Isolation of Malicious Nodes in MANET Secure Routing Protocol with Malicious Nodes Detection for Ad Hoc Networks (ARIADNE) A Cooperative Black hole Node Detection Mechanism for ADHOC Networks Malicious node detection in Ad Hoc networks using timed automata Addressing Collaborative Attacks and Defense in Ad Hoc Wireless Networks dpraodv: a dynamic learning system against black hole attack in aodv based manet and Performance Evaluation of the Impact of Attacks on Mobile Ad hoc Networks . All these research work reveals that a single or to a maximum of two or three

29

attacks identication and prevention using some approach is considered. Our solution to this research gap is to provide a semantic security scheme that considers a minimum of 4 attacks identication and prevention parallelly using a simple acknowledgement approach. The above related study justies that, the proposed scheme is not considered anywhere and is a new security solution for network layer attacks. The reason to concentrate on network layer attacks because; as we know a MANETS network connectivity is mainly through the link-layer protocols and network-layer protocols. The Link-layer protocols are used to ensure one-hop connectivity while network-layer protocols extend this connectivity to multiple hops [2]. So only to incorporate MANETS security we can consider two possible counter measures namely, link-layer security and network-layer security. Link-layer security is to protect the one-hop connectivity between two adjacent nodes that are within each others communication range through secure protocols, such like the IEEE 802.11 WEP protocol [3] or the more recently proposed 802.11i/WPA protocol . The network-layer security mainly considers for delivering the packets between mobile nodes in a secure manner through multihop ad hoc forwarding. This ensures that the routing message exchange within the packets between nodes is consistent with the protocol specication. Even the packet forwarding of every node is consistent with its routing states. Accordingly, the protocols are broadly classied in to two categories: secure ad hoc routing protocols and secure packet forwarding protocols. The paper mainly discusses about the

network-layer security. 4.3.4 PROPOSED APPROACH

The routing attacks like black hole, gray hole, worm hole, rushing attack, DOS attack, ooding etc; can become hazardous to the network-layer protocol which needs to be protected. Further the malicious nodes may deny forwarding packets properly even they have found to be genuine during the routing discovery phase. A malicious node can pretend to join the routing correctly but later goes on ignoring all the packets that pass through it rather than forwarding them. This attack is called black hole, or selective forward of some packets is known as grey hole attack. The basic solution needed to resolve these types of problems is to make sure that every node in a network forwards packets to its destination properly. To ensure this kind of security to network layer in MANETS a new secure approach which uses a simple acknowledgement approach and principle of ow conservation is proposed here. As a part of this research work we have tried the same. 4.3.5 CONCLUSION AND FUTURE WORK

In mobile ad hoc networks, protecting the network layer from attacks is an important research topic in wireless security. This paper describes a robust scheme for networklayer security solution in ad hoc networks, which protects both, routing and packet forwarding functionalities without the context of any data forwarding protocol. This ap-

30

proach tackles the issue in an ecient manner since four attacks have been identied parallelly. The overall idea of this algorithm is to detect malicious nodes launching attacks and misbehaving links to prevent them from communication network. This work explores a robust and a very simple idea, which can be implemented and tested in future for more number of attacks, by increasing the number of nodes in the network. To this end, we have presented an approach, a networklayer security solution against attacks that protects routing and forwarding operations in the network. As a potential direction for future work, we are considering measurement of more number of network parameters, to analyze the performance of such a network using the proposed approach.

all the packets passing through them. In this paper a study that will through light on such attacks in MANETS is presented. The paper also focuses on dierent security aspects of network layer and discusses the eect of the attacks in detail through a survey of approaches used for security purpose. 4.4.2 INTRODUCTION

4.4

Network Layer Attacks and Defense Mechanisms in MANETS- A Survey

ABSTRACT

4.4.1

The foremost concerned security issue in mobile ad hoc networks is to protect the network layer from malicious attacks, thereby identifying and preventing malicious nodes. A unied security solution is in very much need for such networks to protect both route and data forwarding operations in the network layer. Without any appropriate security solution, the malicious nodes in the network can readily act to function as routers. This will solely disturb the network operation from correct delivering of the packets, like the malicious nodes can give stale routing updates or drop

Mobile ad hoc network (MANET) is a type of wireless ad hoc network, and is a selfconguring network of mobile devices connected by any number of wireless links. Each device in a MANET is free to move independently in any direction, and will therefore change its links to other devices frequently. Each must forward trac unrelated to its own use, and therefore be a router. The primary challenge in building a MANET is equipping each device to continuously maintain the information required to properly route trac. Such networks may operate by themselves or may be connected to the larger Internet. Many academic papers evaluate protocols and abilities assuming varying degrees of mobility within a bounded space, usually with all nodes within a few hops of each other and usually with nodes sending data at a constant rate, packet drop rate, the overhead introduced by the routing protocol, and other measures. Security is an essential service for wireless network communications. However, the characteristics of MANETS pose both challenges and opportunities in achieving security goals, such as condentiality, authentication, integrity, availability, access control, and nonrepudia-

31

tion [1]. The countermeasures can be considered as features or functions that reduce or eliminate security vulnerabilities and attacks. First, in this paper an overview of network layer attacks is given, and then the security counter measures. Since in MANETS the nodes dynamically set up paths among themselves to transmit the packets, it is referred as infrastructure less network. The nodes in MANET can communicate directly if they are in within each others wireless transmission ranges otherwise they have to rely on some other nodes to transmit messages if the nodes are outside each others transmission range [2]. Thus, several intermediate hosts relay the packets which are sent by the source host before they reach the destination host, which in turn leads to a multi-hop scenario I.e. each node, will act as a router. The nodes cooperation is very much important for a successful communication. Thus, a MANET has several salient characteristics [3]: dynamic topologies, resource constraints, limited physical security, and no infrastructure. Possible applications of MANET include: Soldiers relaying information for situational awareness on the battleeld, business associates sharing information during a meeting; attendees using laptop computers to participate in an interactive conference; and emergency disaster relief personnel coordinating eorts after a re, hurricane, or earthquake [1]. The other possible applications [2] include personal area and home networking, location-based services, and sensor networks. There are a wide variety of attacks that target the weakness of MANETS. For example, routing messages are an impor-

tant component of mobile network communications, as each packet needs to be passed quickly through intermediate nodes, which the packet must traverse from a source to the destination. Malicious routing attacks can target the routing discovery or maintenance phase by not following the specications of the routing protocols. There are also attacks that target some particular routing protocols, such as DSR, or AODV [4] [5]. More sophisticated and subtle routing attacks have been identied in recent published papers, such as the black hole (or sinkhole) [6], Byzantine [7], and wormhole [8] [9] attacks. Currently routing security is one of the hottest research areas in MANET, so only the research initiative is taken for a specic layer like network layer in OSI model [1]. This paper is organized as follows. In Section 2, description about the network layer attacks is given. In Section 3, proposed solutions for the dierent network layer attacks are discussed, including multilayer attacks. In section 4, a discussion on open challenges and future directions is given. 4.4.3 NETWORK SECURITY ATTACKS

The connectivity of mobile nodes over a wireless link in MANETS which is multihop in nature strongly relies on the fact that ensures cooperation among the nodes in the network. Since network layer protocols forms connectivity from one hop neighbors to all other nodes in MANET, the assurance of cooperation among nodes is required. Recently variety of network layer targeted attacks have been identied and heavily studied in re-

32

search papers. As a consequence of attacking network layer routing protocols, adversaries can easily disturb and absorb network trac, inject themselves into the selected datatransmission path between the source and destination, and thuscontrol the network trac ow, as shown in Figure 1, where a malicious node M can interfere itself in between any of the intermediate nodes participating in the communication in the chosen path (in the gure 1 to N represents the number of intermediate nodes) between source S and destination D [1]. 4.4.4 Network Layer Attacks Description

neighbors to initiate a RREQ packet.

4.4.5

DEFENSE AGAINST NETWORK LAYER ATTACKS

Black hole Attack: In routing mechanism of ad hoc networks three layers namely physical, MAC and network layers plays a major role. As MANETs are more vulnerable to various attacks, all these three layers suer from such attacks and cause routing disorders. The variety of attacks in the network layer diers such as not forwarding the packets or adding and modifying some parameters of routing messages; such as sequence number and hop count. The most basic attack executed by the nodes in the network layer is that an adversary can stop forwarding the data packets. The consequence caused by this is that, whenever the adversary is selected as an intermediate node in the selected route, it denies the communication to take place. Most of the times the black hole attack is launched by the adversaries, whenever AODV is used as the data forwarding protocol. Consider a malicious node which keeps waiting for its

The previous section reveals the possibility of various attacks on the network layer and now the focus is on the several security measures taken to overcome these attacks. As it is a known fact cryptography is one of the most common and reliable means to ensure security in MANETS. The main notions for cryptography are condentiality, integrity, authentication and non-repudiation. The cryptography is discussed in detailed in [17]. MANETS have certain challenges in key management due to lack of infrastructure, absence of dedicated routers and mobility of nodes, limited processing power and limitation of battery power, bandwidth and memory. The main requirement to ensure security 1in MANETS is to have a secure routing protocol which should have properties to detect malicious nodes, guarantee of exact route discovery process, maintaining condential network topological information and to be selfstable against attacks. SAR (Secure-Aware Ad Hoc Routing protocol), which denes a level of trust as a metric for routing and as an attribute for security for routing. SAR using AODV uses encryption and decryption process using a common key [18]. The main drawback with SAR protocol is whenever the levels of security rise; it needs dierent keys for dierent levels, thereby increasing the number of keys

33

4.4.6

OPEN CHALLENGES AND of PGP (Pretty Good Privacy). Symmetric FUTURE DIRECTIONS cryptography h computation eciency, yet it suers from potential attacks on key agreeSecurity in MANETS is such a hot topic ment or key distribution. For example, the among the research communities, if it is as- Die-Hellman (DH) scheme is vulnerable to sured properly it can be used as a success the man-in-the-middle attack. Many complifactor and for the widespread deployment of cated key exchange or distribution protocols the network. Several types of attacks in net- have been designed, but for MANET, they work layers have been identied and analyzed are restricted by a nodes available resources, recently in most research papers. Security dynamic network topology, and limited bandcountermeasures and the defense against for width. Ecient key agreement and distrieach of the network attacks so far designed bution in MANET is an ongoing research and implemented for MANETS are presented area. Most of the current work is on prein the above sections. The research propos- ventive methods with intrusion detection as als till date, in MANETS are based upon a the second line of defense [1]. One interestspecic attack. They could work well in the ing research issue is to build a mechanism presence of designated attacks, but there are which uses many approaches together withmany unanticipated or combined attacks that out the use of key management to ensure remain undiscovered. A lot of research is still more level of security in MANET. Building on the way to identify new threats and create a sound robust semantic security approach secure mechanisms to counter those threats. and integrating it into the current prevenMore research can be done on the robust tive methods can be done in future research. key management system, trust-based proto- Since most attacks are unpredictable, a recols, integrated approaches to routing secu- siliencyoriented security solution will be more rity, and data security at network layer. Here useful, which depends on a multi-fence secuare some research topics and future work in rity solution. less network. The nodes in the area: a) Cryptography is the fundamen- MANET can communicate directly if they tal security technique used in almost all as- are in within each others wireless transmispects of security. The strength of any crypto- sion ranges otherwise they have to rely on graphic system depends on proper key man- some other nodes to transmit messages if agement. The public-key cryptography ap- the nodes are outside each others transmisproach relies on the centralized CA (certi- sion range [2]. Thus, several intermediate fying authority) entity, which is a security hosts relay the packets which are sent by the weak point in MANET. Some papers pro- source host before they reach the destination pose to distribute CA functionality to mul- host, which in turn leads to a multi-hop scetiple or all network entities based on a se- nario I.e. each node, will act as a router. cret sharing scheme, while some suggest a The nodes cooperation is very much imporfully distributed trust model, in the style tant for a successful communication. Thus, 34

a MANET has several salient characteristics [3]: dynamic topologies, resource constraints, limited physical security, and no infrastructure. Possible applications of MANET include.

TCP/IP MANET

OVER

bodes well for a future where TCP and its excellent qualities is part of extending wirebased services into the wireless multi-hop domain. This report focuses on the evolution and adaption of the TCP protocol. There are other protocols that aim to replace TCP entirely, for instance through implementing ow and congestion control on top of the User Datagram Protocol (UDP). These solutions are beyond the scope of this report.

The Transmission Control Protocol (TCP) has been, and continues to be, an essential protocol for Internet communication. Without its rate control, trac congestion would have rendered the Internet useless. However, TCP makes several assumptions about the network. It assumes that network congestion, and not transmission errors, causes packet loss. It also assumes that the Round Trip Time (RTT) is relatively constant (little jitter) and that rerouting happens very quickly. None of these assumptions are easily satised in MANETs, which results in TCP having substantial problems when employed in such environments. TCP has been improved several times after its rst version in 1981. In recent years, the focus has mainly been on optimizations due to the ever-increasing link capacity of wired networks. These proposals are not directly useful for the employment of TCP in MANETs. However, there has become more interest during the last decade in improving TCP for communication in multihop wireless networks, although the main TCP research activity still has been focused at High Performance Computing (HPC). The increased research eort on TCP in MANETs

5.1
5.1.1

TCP in MANETs lenges and Solutions

Introduction

chal-

Military communication on a tactical level is becoming IP-based. This allows the employment of one common communication infrastructure for multiple systems, enabling the network-based defense paradigm. With IP-based connectivity comes also a desire to interconnect wired and wireless communications systems. There is an expectation that services used in wired networks today also will be available in the wireless domain. In the wireless domain, cellphone technology has shown users that IPbased web communication is feasible. However, this communication technology requires infrastructure in the form of a highcapacity backbone1 network and one hop wireless communication between the client terminal and base stations that connect to the backbone. Current wireless communication in the military tactical domain consists mainly of point-to-point radio links and one-hop broadcast voice=Situational Aware-

35

ness (SA) data. However, there is a lot of ongoing work focusing on interconnecting the various radio systems using Mobile Ad hoc NETwork (MANET) technology, to create heterogeneous MANETs. MANETs are self-conguring infrastructure-less networks that adapt dynamically to changing environments. In contrast to cellphone technology, MANETs are able to support multi-hop wireless communication over a shared medium. However, the capacity and performance of MANETs are much lower, compared to cellphone networks, and informing future users and service developers on the limitations as well as the advantages of this technology is essential for proliferation of the MANET technology. While MANET technology is very suitable for tactical communication, many IP-based protocols are not directly usable in MANETs. These protocols were developed in a strictly wire-based network domain, where attributes like interference and packet loss are less dominant and better controlled than in wireless multi-hop networks. For instance, queue loss is the sole contributor to packet loss, while medium-based bit errors are but non-existent. In MANETs, the Bit Error Rate (BER) is much higher than in wired networks (several orders of magnitude). Protocols that anticipate the cause of packet loss to be caused by queue tail drop may make the wrong assumption in MANETs, reacting badly in this situation.

OSI network architecture stack [1]. It guarantees that a stream of bytes sent from the sender program on one computer is delivered reliably and in the same order to the receiver program on the other computer. The counterpart to the reliable TCP service is the User Datagram Protocol (UDP), which provides a datagram service where latency is reduced at the cost of data delivery reliability. A few key features set TCP apart from UDP: Ordered data transfer. Retransmission of lost packets. Error-free data transfer. Flow control. Congestion control. This report gives only a brief introduction to the functions of the TCP protocol, to give the reader an understanding of the basic functions in the TCP protocol, and the dierences between dierent TCP variants. If more information is desired, this can be acquired through several sources, including the many RFCs (introduced in Chapter 2.3.1) that describe the functions of TCP formally, three innovative books covering TCP/IP by W. Richard Stevens and also more easily digestive works such as P. D. Amers presentation. 5.1.3 SACK TCP

Selective Acknowledgment (SACK) TCP was another improvement to TCP proposed in 1996, in RFC 2018 [10]. The earlier variants of TCP, even back to Tahoe, implement a cumulative acknowledgment scheme where a lost packet results in duplicate ACKs for each subsequently received packet. Re5.1.2 TCP in wired networks lying purely on the cumulative acknowledgTCP is a transport protocol that provides a ment scheme can lead to ineciencies when number of services for higher layers in the packets are lost. An example of this is a 36

case where 10,000 bytes are sent in 10 different TCP packets, and the rst packet is lost during transmission. Using the cumulative acknowledgment scheme, the receiver cannot say that it received the bytes 1,000 to 9,999 successfully, and only failed to receive the rst packet, containing the bytes 0 to 999. Thus the sender may then have to resend all 10,000 bytes. To solve the inecient retransmission problem, TCP may employ the SACK option7, which allows the receiver to acknowledge discontinuous blocks of packets that were received correctly. This is an additional mechanism to the sequence number of the last contiguous byte received successively, as in the basic TCP acknowledgment. The acknowledgment can specify a number of SACK blocks, where each SACK block is conveyed by the starting and ending sequence numbers of a contiguous range that the receiver correctly received. In the example above, the receiver would send SACK with sequence numbers 1,000 and 9,999, and the sender will therefore only retransmit the rst packet, bytes 0 to 999. 5.1.4 Challenges MANETs for TCP in

The challenges for TCP in MANETs span all the layers below the transport layer in the OSI network stack [1]. At the PHY layer, interference and fading may result in bit errors and lost packets. At the MAC layer, the medium access may induce delay and is not able to totally avoid collisions, potentially causing packet loss if retransmission mechanisms are unable to salvage the prob-

lem. Retransmissions will also create delay and jitter. Some Medium Access Control (MAC) protocol implementations are able to dynamically change the data rate based on the transmission success. At the network layer, the routing protocols delay in detecting topology changes may lead to periods without connectivity. Also, the end-to-end transmission time will change as a result of changing paths between the source and destination. The IEEE 802.11 wireless stack [34, 35] is by far the most common wireless platform that is used for ad hoc networking today. Many MANET challenges have been identied based on work with the 802.11 platform, and in some cases the problems and subsequent solutions have focused more on mending the 802.11 standard than on addressing MANET problems generically. However, it is important to note that the 802.11 stack implements several mechanisms that are necessary for a functional MANET based on a Carrier Sense Multiple Access (CSMA)/Collision Avoidance (CA) MAC protocol. The functionality of 802.11 as a MANET carrier has been studied extensively, and in this TCP for MANET memo, the examination by Xu and Saadawi is particularly interesting. In 2001, Xu and Saadawi [36] examined how well or rather how badly TCP was supported in an IEEE 802.11 MAC MANET, with the focus on showing why the 802.11 protocol was unt for MANET communication.

37

5.1.5

Solutions to improve TCPs per- tions that may be considered compatible with formance current TCP implementations, through solutions that changes the behavior of one or both Several proposals to mend the many chal- the end-points, to solutions that require all lenges encountered by TCP in MANETs have nodes in the network to implement changes been generated through research. Some of to support the solution: Gateway-oriented these solutions are presented below. The in- Changes limited to the source and/or the destention is to give the reader an overview of tination. These are further grouped accordthe types of solutions that have been brought ing to compatibility: forward, for reference and for better understanding of the ways that TCPs challenges 5.1.6 Conclusions can be met. IETF has addressed TCPs challenges through several RFCs. Two of This report has presented the functionality of them are RFC3135 and RFC3449. RFC3135 TCP, in addition to its development for wired is a survey from 2001 of Performance En- networks as the congestion window mechahancing Proxys (PEPs) employed to im- nism was invented and rened. The chalprove degraded TCP performance caused lenges that TCP faces in MANETs have been by characteristics of specic link environ- explained from a OSI network stack perspecments. RFC3449 presents best current prac- tive. After this, a large number of diertices (from 2002) with regards to network ent proposals to make TCP work better in path asymmetry. There are a number of MANETs were presented and discussed. The surveys delving into the challenges of TCP discussion of the ndings showed the comin MANETs and possible solutions. Wang plexity of the solutions and the problem of and Zhang presents a survey on TCP over choosing one single solution to improve the MANETs, introducing three major chal- TCP performance, and more research is deflenges for TCP. Two other publications that initely needed before any solution could be look at TCP and congestion control for deployed as part of an operational network. MANETs are]. Al Hanbali et al. present in With the ndings of this report taken into a survey of TCP alternatives for MANETs, consideration, it is legitimate to ask whether classifying the alternatives in cross-layer and TCP is the only choice for reliable communilayered proposals. A very thorough survey cation in MANETs. There are eorts focusof TCP and similar congestion control proto- ing on UDP-based communications that may cols for MANETs is presented in. In the fol- handle this communication well enough, emlowing subchapters, many solutions for TCP ploying Automatic RepeatreQuest (ARQ) at in MANETs are presented. The solutions the MAC-layer and handling reliability at the are grouped according to the solutions re- application layer. In such a case, the congesquirement for changes, spanning from the tion control oered by TCP must be handled Gateway (GW)- oriented low impact solu- with other mechanisms, through admission 38

control and local Quality of Service (QoS). Another direction is within the research of Information Centric Networking (ICN), where congestion avoidance is also addressed.

5.2

PERFORMANCE EVALUATION OF TCP/IP PROTOCOL

Transmission control protocol (TCP) is a transport protocol designed specically for wired Internet. In wireless ad hoc network, changes in topology can occur frequently and unpredictably which leads to packet loss and delay. TCP misinterprets that condition as congestion and it reacts by reducing the transmission rate and causes the performance to degrade and lower throughput. The TCP has to be modied if to recognize the dierence between packet loss due to link failure or congestion. To adapt TCP to this demanding paradigm, some modications have been proposed. This study evaluates the performance of the proposed enhanced TCP for mobile ad hoc network (MANET) environment. The enhanced TCP is integrated with Snoop protocol and installed in every node of the system model. The system model is designed and developed using OPNET software design tool. The simulation results show the enhanced TCP produces improved performance with higher throughput and support for node 5.2.2 mobility. 5.2.1 INTRODUCTION

nodes that are capable of communicating among themselves without the use any communication infrastructure. Mobile users can roam in the specied area and still can communicate directly with other mobile node. For this purpose, every mobile user also serves as a relay or router for other nodes. Previously most researchers are interested in routing protocol challenges for wireless ad hoc networks. However in this study the focus is on developing a mechanism, which can improve the performance of transmission control protocol (TCP) [1] along with Internet protocol (IP) [2] for MANET. The TCP/IP is widely accepted in Internet as thereliable end-to-end transport protocol. While the IP handles the actual data delivery, the TCP manages the individual packet for ecient routing in the Internet. By using ow control, sequence number, acknowledgement and timer, the TCP can ensure delivery from the sending process to the receiving process correctly, orderly and error-free. However in MANET environment, network topology changes and discontinuities occurs frequently causing packet loss, which TCP misinterprets as congestion [3]. The TCP will react by reducing the transmission rate and aect the performance to degrade with lower throughput. CURRENT NISM TCP MECHA-

In this section, the more popular proposed schemes to improve TCP performance for Over the past few years, mobile ad hoc MANET architecture are presented. networks (MANET) are regarded as mobile I. Snoop Protocol Snoop protocol [7] is de39

signed to be TCP aware. The snoop protocol modies the network layer software on the base station to cache the TCP packets and perform local retransmissions across the wireless links. It is implemented as a layer in TCP/IP architecture stack. It can be located both at the access point and the mobile node. The Snoop protocol runs on a snoop agent that is implemented in the base station or wireless devise. The agent monitors the packet that passes through and caches the packet into a table for TCP connection. Then the agent forwards the packets to their destinations and monitors the corresponding acknowledgements (ACKs) without forwarding the ACKs to the sender. II. Mobile-TCP Another solution is the mobile-TCP [8], which is a transport protocol for mobile computing designed specically for lengthy disconnections or frequent disconnections. Mobile-TCP informs the sender that a disconnection has occurred. It ensures that hando is ecient and the end-to-end TCP semantics are maintained. If TCP sender detects a packet loss (duplicate acknowledgement or timeout) it will perform retransmissions but without reducing its window size. Once disconnection ends, the sender is informed to resume normal operation. III ATCP Protocol While the ATCP protocol [9] utilizes the network layer feedback. It does not impose changes to the standard TCP itself. It is a layer between IP and TCP that follows the network state information of Destination Unreachable message and the Explicit Congestion Notication (ECN) message provided by ICMP. In case of route failure, the ICMP Destination Unreachable

message will allow the TCP sender to enter persist state. The ECN message can dierentiates packet loss due to congestion or wireless transmission errors. In case of loss, the TCP sender enters persist state and in case of congestion, ATCP will not interfere with TCP congestion control algorithms but forwards the packet to TCP so that it can invoke its congestion control mechanism 5.2.3 ENHANCINGCURRENTTCP

The criteria or factors considered in this study in improving the current TCP to meet the demand of node mobility in MANET are identied. There are two major factors considered in order to improve the TCP throughput. I. Route Failure: The protocol must be able to distinguish between packet loss due to route failure and packet loss due to congestion. II. Congestion Window: The protocol must not use the old route congestion window size for the new route, since it is unlikely for similar conditions also continue to exist in the new route. In reality, some of the ideas developed for the current solutions may also be enhanced and new paradigms can be designed. In this work, we propose to integrate the current TCP with the Snoop protocol to match our requirement. The following factors are relevant to the proposed enhance wireless TCP protocol: The new protocol must preserve the TCP end-toend acknowledgement and semantic in order to identify the presence of various network condition. The protocol must support wireless transmission er-

40

rors, route failures and handos. The protocol can handle congestion well. The protocol uses a new congestion window size for a new route. The protocol will not aect the TCP structure. III.Snooping in Ad Hoc Network: To improve the TCP performance the Snoop protocol is integrated in the ad hoc network, where each node acts as host as well as router for other nodes. In other words, every node in the network is equipped with a Snoop module. Snooping at each node will increase the packet availability even when nodes rearrange themselves by joining or leaving the network. The packet is sent from one mobile host to another with random speed and to random destinations. If two hosts are close enough to each other they can communicate directly. However, if they are far apart, then other host can relay the packets to and from these two distant users during the communication. Working from the original Snoop protocol, all modications take place at the base station and the hosts. The dierence between wired network and ad hoc network is in the degree of the physical infrastructure present. The original Snoop protocol was designed with the vision of wireless network becomes an extension to the wired network. In ad hoc network scenario, each mobile node works with other nodes so as to dynamically maintain network topology by broadcasting packets to all the neighboring nodes. Each mobile user serves as a relay or a router for other nodes and allows communication among them. IV.Snoop Cache: The Snoop cache is used to cache packets that are received from the higher layer. It will retransmit these packets

once Snoop detects a loss packet due to the received of duplicate ACK. As mention previously,the Snoop in each node will broadcast the packets toall neighboring nodes. Caching the packets at all nodes will increase the packets availability. Thus, each node requires bigger cache size. The number of packets in the cache will decrease as the Snoop protocol receives the ACK. The number of cached packets will become zero when the TCP connection isthe Snoop in each node will broadcast the packets to all neighboring nodes. Caching the packets at all nodes will increase the packets availability. Thus, each node requires bigger cache size. The number of packets in the cache will decrease as the Snoop protocol receives the ACK. The number of cached packets will become zero when the TCP connection is completed. 5.2.4 CONCLUSIONS

The result obtained is validated since there is small dierence between the two throughputs, theoretical and measured value. As the number of hops increases, the throughput degrades due to factors such as interference and high bit error rate. That is why the performance of trac packet decreases as the data transmission distance from sender to receiver increases. From the performance graphs shown, the Snoop protocol can improves the TCP throughput performance without changing other layer stacks. The TCP/IP layer is unmodied, so the semantics is preserved. The throughput performance is improved with Snoop protocol. Nodes with Snoop protocol can transfer

41

more data packets compared to those without Snoop when transmitting almost the same data packets (sequence number) through the network. In addition, the result also shows that Snoop works well in preventing TCP congestion window from shrinking, hence increases the performance signicantly. This proves the eect of caching at each node produce better results. Therefore, the assumption that all losses are due to congestion becomes quite problematic over wireless links. Packets losses are mostly due to the bit errors caused by the eects of environmental impact such as noise and delay. Thus, numerous proposals for appropriate TCP modications existed. To this end, we have implemented a new improved TCP protocol in the OPNET environment using the schemes of Snoop protocol, which serves as a benchmark in this study and analysis. Snoop at every nodes produce an improvement in the throughput. This mechanism seems appropriate and leads to the implementations of TCP over wireless link with better performance. The results obtained from extensive simulation run for mobile ad hoc network proves the new wireless TCP produces better throughput performance.

suer from severe performance degradation. To handle a packet loss, conventional TCP retransmits the lost packet from its source. However, when error rate is high, it may have to take several retransmissions to deliver a packet to its destination successfully. Considering that most applications on a MANET prefer shorter packet delivery time to higher bandwidth, this paper proposes a Hop-byHop TCP protocol for MANET aiming to accelerate packet delivery. Hop-by-Hop TCP makes every intermediate node in the transmission path execute a local TCP to guarantee the transmission of each packet on each link. The retransmission of a lost packet is right at the transmitting end of the link where the packet is lost. It takes less time in average to deliver a packet in a high error rate environment. The performance of our approach is evaluated by simulation using NS-2 simulator. Our experiments show that our proposed protocol outperforms TCP Reno by 25.7 5.3.1 INTRODUCTION

5.3

Hop-by-Hop MANET

TCP

over

Abstract-In a MANET environment, communication links are unstable due to various reasons. Error rate is higher and bandwidth is smaller than xed networks. Running conventional TCP protocol over MANET will

A Mobile Ad hoc Network (MANET) is composed of a group of mobile computing devices (nodes) that are equipped with Wireless LAN (WLAN) capability. Nodes can transmit packets to each other to construct an Intranet without the assistance of any base station (Access Point). Nodes in a multi-hop MANET help each other to forward packets from hop to hop such that two nodes that cant hear each other can transmit data to each other. In this way, the connectivity of a MANET is greatly enhanced. In a

42

MANET environment, communication links are unstable due to various reasons such as interference of radio signal, radio channel contention and mobility of nodes. Furthermore, the multi-hop feature increases the data trafc signicantly that in turn increases error rate. In summary, in a MANET, error rate is higher and bandwidth is smaller as compare to xed networks. As a consequence, running conventional TCP protocol over a MANET will suer from severe performance degradation. To handle a packet loss, conventional TCP retransmits the lost packet from its source. However, when error rate is high, it may have to take several retransmissions to deliver a packet to its destination successfully. As a result, the eective bandwidth will be much lower and the average packet delivery time will be much longer. Furthermore, packet losses may also activate TCPs congestion control causing too many what is so called slow start that will further impair the performance of packet delivery. As a result, the eective bandwidth is much lower and the average packet delivery time will be much longer. In fact, most applications on a MANET prefer shorter packet delivery time to higher bandwidth. However, most versions of TCP are all designed to achieve higher bandwidth, not shorter packet delivery time. Therefore, it would be a good idea to redesign a TCP by trading bandwidth for faster packet delivery. Redesigning TCP may not be easy to implement on a WAN (Wide Area Network) because upgrading a large number of routers in a WAN is almost a business impossible. However, MANET has no such concern so that it is easy for a MANET to embrace

any new approach. This paper proposes a Hop-by-Hop TCP protocol aiming to accelerate packet delivery. Hop-by-Hop TCP makes every intermediate node in the transmission path execute a local TCP to guarantee the transmission of each packet on each link. The retransmission of a lost packet is right at the sender end of the link where the packet is lost. It doesnt have to retransmit a lost packet all the way from its source node. It takes less time in average to transmit a packet to its destination in a high error rate environment. 5.3.2 RELATED WORK

TCP is a connection-oriented reliable data delivery protocol. It provides a reliable transport service between pairs of hostsusing the network layer service provided by the IP protocol. Whereas the IP protocol deals only with packets, TCP enables two hosts to establish a connection and exchange streams of data. TCP guarantees the delivery of data and also guarantees that packets will be delivered in the same order in which they were sent.Packets may be lost in the transmission path due to various reasons such as network congestion and radio channel error. TCP adopts a complicate protocol to guarantee the delivery of packets. TCP modules reside at the both ends of a connection which may have quite a few intermediate nodes in between. Thus, the source node of a TCP connection must determine appropriate data rates based on its own poor knowledge of network status. Because most hosts on the network do not have a good knowledge of the network status, it is impossible for them to have a perfect

43

data rate control, network will be congested from time to time. Network elements including routers and end terminals must work hard to avoid network congestion. The congestion control within a TCP plays a critical role in adjusting data rate to reduce network congestion. Based on some window-adjustment algorithm, a TCP not only guarantees the successful packet delivery, but also maintains an appropriate data rate. The two indicators of network status are packet traveling time and the success/failure of package delivery. Therefore, most TCP variants count on these indicators to guess (estimate) the available bandwidth over the packet delivery path and to adjust data rate accordingly. The accuracy and the promptness of bandwidth estimation are dependent on many factors such as trac stability and path length. Not surprisingly, most TCP variants are suering from some performance shortcomings, unnecessary network congestion and slow start. Unfortunately, a MANET is generally slow and unstable such that running TCP over MANET will suer from even severe performance problems. In summary, all TCP variants that are designed for conventional networks may not be appropriate for MANET. There is a need to redesign TCP protocol for MANET to improve its performance. 5.3.3 End-to-end TCP

1. Instead of interacting with IP Layer, the sender module forwards packets to the One-Hop TCP module, and the receiver module receives packets from the One-Hop TCP module. One-Hop TCP in each node forwards data packets hop by hop to the destination node. Similarly, ACK packets for End-to-End TCP, called End-to-End ACKs, are forwarded to the source node using OneHop TCP in the opposite direction. 2. Set a maximum threshold on the size of CWND to prevent it from over growth. (As compared to the conventional WAN environments, it is much easier to estimate the available bandwidth for a MANET. Setting a threshold can reduce congestions.) 3. Set a larger initial RTO (Retransmission Time Out) value. 5.3.4 3.2 One-Hop TCP

To avoid reinventing the wheel, we reuse an existing popular TCP protocol, NewReno, for the End-to-End TCP with severalmodications. In fact, it can be replaced with any other version easily.

One-Hop TCP is a light-weight version of TCP running on each node to forward received packets to the next node packet by packet reliably. Major modications are as follows: 1. add the IP address of current node to the packet header such that the receiver knows where to send Local ACK; 2. set the local RTO based on link characteristics; 3. Set CWND to 1 4. remove all CWND adjustment mechanism; 5. set the upper threshold for the number of retransmissions. Since the speed of One-Hop TCP is very critical to the performance of the entire TCP, it must be very ecient. Many TCP features, such as packetization and congestion control, are removed from One-Hop TCP. It is more

44

like an enhancement to the link layer protocol to accommodate link failure. Furthermore, as mentioned earlier, only one session of OneHop TCP is sucient to manage all End-toEnd TCP sessions that pass through the same link. Many overhead can be saved. CWND is set to 1 for two reasons. First, the main reason that the sliding window mechanism is used in conventional TCPs is to allow more than one packet pending in the intermediate router. However, there is no such need for a one-hop link. Furthermore, if next packet were allowed to transmit before the ACK of the previously transmitted packet is received, the transmissions of the two packets will have to compete to each other for the radio channel. Thus, CWND is set to 1. In other words, after transmitting a packet, the sender module has to wait for the ACK packet before it can transmit next packet. In this way, complicated CWND size adjusting mechanisms become useless in One-op TCP and are all removed. 5.3.5 CONCLUDING

protocol outperforms TCP Reno by 25.7

TRANSPORT LAYER IN MANET

In this paper, we propose a Hop-by-Hop TCP protocol that makes every intermediate node in the transmission path of a TCP execute a local TCP (One-Hop TCP) to guarantee the transmission of each packet on each link. The retransmission of a lost packet is right at the transmitting end of the link where the packet is lost. It takes less time in average to deliver a packet in a high error rate environment. The performance of our approach is evaluated by simulation using NS-2 simulator. Our experiments show that our proposed

Millions of people nowadays have portable computers and they generally want to read their e-mail and access their normal le systems wherever in the world they may be. This demand for mobility has fueled the rapid progression of computer communication technologies from networks consisting of both stationary hosts routers, to networks consisting of mobile hosts and stationary routers, and more recently to the other extreme case of networks having both mobile hosts and mobile routers. This last case of infrastructure-less networks of mobile hosts and mobile routers are called ad hoc networks or MANETs (Mobile Ad Hoc NETworks). It has been claimed that MANETs are a fundamentally awed architecture. This is because Mobile Ad Hoc networks are almost never used in practice and almost every wireless network nodes communicate to base-stations and access points, instead of co-operating to forward packets hopbyhop. In this paper, we take the position that MANETs cannot be declared a fundamentally awed architecture. The reasoning behind this position is that MANETs are still an emerging technology that has received intensive attention only recently. All aspects of the MANET concept are yet to be identied and fully understood. Research on MANETs is still in its early stages and considerable research eort

45

is still necessary.

6.1

INTRODUCTION

In the near future, a truly pervasive computing environment can be expected with traditional home appliances attached with computing communicating powers and small devices like mobile phones, Personal Digital Assistants wearable computers enhancing information processing and accessing capabilities with mobility. The MANET technology truly supports pervasive computing because in many contexts information exchange between mobile units cannot rely on any xed network infrastructure but on rapid conguration of a temporary wireless network. 1. Military, Industrial and Commercial applications involving cooperative mobile data exchange. 2. Inexpensive alternatives or enhancements to cell-based mobile network infrastructures. 3. Future military networking for robust, IP-compliant data services within mobile wireless communication networks consisting of highly dynamic autonomous topology. 4. With satellite-based information delivery, MANET can be used for re/ safety/rescue operations or other scenarios requiring rapidlydeployable communications with survivable,ecient dynamic networking. The principle behind MANET is multihoprelaying, which is nothing new as it traces its roots back to 500 B.C. Darius I, the king of Persia devised an innovative communication system based on this principle. He placed a line of shouting men positioned on tall structures and heights and this use of ad hoc

communication proved to be 25 times faster than normal messengers at that time. As a technology for dynamic wireless networks, Ad hoc networking has been deployed in military since 1970s. Commercial interest in such networks has recently grown due to the advances in wireless communications. A new working group for MANET has been formed within the Internet Engineering Task Force (IETF), aiming to investigate and develop a framework for running IP based protocols in ad hoc networks. 6.1.1 The MANET Technology

A MANET is a collection of mobile platforms (e.g., a router with multiple hosts and wireless communications devices) herein simply referred to as nodes that can dynamically be set up anywhere and anytime without using any pre-existing network infrastructure. The nodes may be located in or on airplanes, ships, trucks, cars, perhaps even on people or very small devices, and there may be multiple hosts per router. It is an autonomous system in which mobile nodes connected by wireless links are free to move randomly. The system may operate in isolation, or may have gateways to and interface with a xed network. In the later operational mode, it is typically envisioned to operate as a stub network connecting to a xed internetwork. 6.1.2 MANET Architecture

The nodes in a MANET can be classied by their capabilities. A Client or Small Mobile Host (SMH) is a node with reduced pro-

46

cessing, storage, communication, and power resources. A Server or Large Mobile Host (LMH) is a node having a larger share of resources. Servers, due to their larger capacity contain the complete DBMS and bear the primary responsibility for data broadcast and satisfying client queries. Clients typically have sucient resources to cache portions of the database as well as storing some DBMS query and processing modules. In a MANET, each node has an area of inuence. This is the area over which its transmissions can be heard. A LMH will initially have a larger area of inuence as it generally has a more powerful battery. As the power level decreases, the area of inuence of any node will shrink because the power available to broadcast is reduced. Network nodes may operate in any of three modes that are designed to facilitate the reduction in power used: I. Active Mode (or Transmit Mode): This is the mode using the most power. It allows both the transmission and reception ofmessages. II. Doze Mode (or Receive Mode): The CPU is capable of processing information and is also capable of receiving notication of messages from other nodes and listening to broadcasts. III. Sleep Mode (or Standby Mode): The CPU does no processing and the node has no ability to send/receive messages. The node is inactive. This mode allows a node to turn itself o for short periods of time without requiring power-up or re-initialization.

6.1.3

Counter claims

The design of MANET involves some very critical dicult issues that have deed a common solution yet and these issues touch all aspects of communication networking. These issues are preventing wide spread use of MANET. So, it can be claimed that MANETs are a fundamentally awed architecture. The following lists the issues that support this claim. 1) Implementation: Traditional MobileNetworks consist of xed routers and mobile hosts while MANETs have mobile routers hosts. This means that MANET is just an extreme case of the Traditional Mobile Network. Ethernet connectors in all the arm rests of an airplane in which mobile computers can plug in is an example of Traditional Mobile Network being successfully implemented and widely used. But there is no such corresponding example for MANET. 2) Traditional Mobile Network as alternative to MANET: MANETs are supposed to provide mobility in communication and this has already been provided by the Traditional Mobile Networks. 3 ) Medium Access: An optimal Medium Access Control protocol for MANET has not been designed yet because of the following issues 6.1.4 Conclusion

The previous section undeniably establishes the fact that MANETs are not a fundamentally awed architecture. Every new technology needs time to evolve mature and

47

MANETs are no exception. MANETs represent a progression of technology, and given a little more time eort will usher in a new era in telecommunications. Condemning MANETs as a fundamentally awed architecture while its still being in its emerging state, amounts to a denial of the human spirit that seeks challenges, remains undaunted by challenges and ultimately triumphs against all odds. A primitive ad hoc MANET-like communication system, as used by Darius I, was sucient to carry the Persian Empire to its largest extent. Similarly, todays Mobile Ad Hoc NETworks (MANETs) will propel our human civilization todizzying heights in the future.

port layer security for authentication, securing end-to-end communications through data encryption and to provide security services for both routing information and data message at network layer. It also handles delay and packet loss. This paper considers military scenarios and evaluate the performance of Securityenhanced-Multicast AODV (Ad hoc On-demand Distance Vector Routing) routing protocol called SNAuth-SPMAODV (Secure Neighbor Authentication Strict Priority Multipath Ad hoc On-demand Distance Vector Routing) with WTLS to minimize the packet dropping by Denial of Service attack. 6.2.1 INTRODUCTION

6.2

Securing Transport using SNAuth-SPMAODV

A mobile ad-hoc network (MANET) is a peer-topeer wireless network where nodes can communicate with each other without any infrastructure. Due to this nature of MANET, it is possible that there could be some malicious and selsh nodes that try to compromise the routing protocol functionality and makes MANET vulnerable to Denial of Service attack in military communication environments. The ultimate goal of the security solutions for MANET is to provide security services, such as authentication, condentiality, integrity, anonymity, and availability to mobile users. To achieve the goals, the security solution should provide complete protection across the entire protocol stack. The primary focus of this work is to provide trans-

In recent years, Mobile ad hoc Networks start gaining attention from the industrial and academic research community due to their wide deployment and inherent nature of solving practical real world applications. Many military and commercial applications have emerged due to the simplicity of the networks and widespread adoption of the technology. Most of the previous ad hoc network researchers have focused on problems such as routing and reliable communication in a trusted environment. However, many applications in reality run only in untrusted environments and secured routing became a challenging one. Applications that may require secure communications include emergency response operations, military or police networks, safety critical business operations such as oil drilling platforms or mining operations. For example, in emergency response operations such as the one after a

48

natural disaster like a ood, tornado, hurricane and earthquake, when regular communication networks are damaged due to natural disasters, emergency rescue teams have to rely upon ad hoc networks for communication.

6.2.3

PROBLEM STATEMENT

6.2.2

SIMULATION MODEL

Using the QualNet network simulator [8], comprehensive simulations are made to evaluate the protocol. Qualnet provides a scalable simulation environment for multi-hop wireless ad hoc networks, with various medium access control protocols such as CSMA and IEEE 802.11. channel and physical layer settings are modied to apply more realistic military scenarios. Note that PRC-999K device is used as a reference model. 802.11 DCF and UDP protocols are used for MAC and a transport protocols, respectively. Also, CBR trac is utilized in the study. As the TCPbased application protocols such as telnet or FTP show unstable performance in mobile wireless communication, it can not evaluate precise performance of routing protocol itself. CBR application model sends one packet per second, which represents relatively low trafc patterns in military environments. Each packet size is 512 Bytes. In military environments, operational network size is very large as compare to conventional case. Nodes in the simulation are assumed to move according to the random way point mobility model. Pause time is xed to 20 seconds. The attackers are positioned around the center of the routing mesh in all experiments.

This research investigates how to integrate security policies of a MANET with secure neighbor authentication that will allow the MANET to function securely in a military environment without degrading network performance. The specic problem to be addressed is how to use secure neighbor authentication of nodes in a multipath routing algorithm in MANET protected from Denial of service attack and provide transport layer security in military environment. Most of such performance analyses are normally done on commercial settings. For instance, wireless LAN technologies in the 2.4 GHz ISM frequency band are generally assumed, oering data rates up to 2 Mbps within the range of 250 m. This paper is motivated by the observation that such propagation and network models assumed by the current ad hoc networking simulations are quite dierent from real world military environments. In fact, a few hundred MHz frequency band (i.e., VHF or even HF) is used with very low data transmission rates (e.g., 384 Kbps) for the military scenarios. Table I summarizes these dierences in terms of a physical layer model. Networking environments such as network size. CONCLUSION Mobile Ad Hoc Network (MANET) is a collection of communication devices or nodes that wish to communicate without any xed infrastructure. Secure, reliable and ecient routing operations in MANET a challenging task. Hence the suitable key management solution clubbed with the routing protocol may be a better option. The primary fo-

49

cus of this work is to provide transport layer security for authentication, securing end-toend communications through data encryption and to provide security services for both routing information and data message at network layer. It also handles delay and packet loss. The proposed approach minimizes the packet dropping by Denial of Service attack (DoS) in the network by applying WTLS in SNAuthSPMAODV routing protocols and compares the results with SNAuth-SPMAODV without WTLS protocols. The simulation results demonstrate the success of the proposed approach and maximize the overall performance of MANET in presence of Denial of Service attack.

tracing on the interaction of TCP with the routing protocol.

6.3.1

INTRODUCTION

6.3

evaluation MANETs

of

TCP

in

Past research eorts have denoted the problematic behaviour of traditional TCP agents in MANET environments and have proposed various remedies across the networking stack. However, there has not been an overall performance evaluation of dierent TCP agents under varying mobility conditions which takes into account past experiences in MANET evaluation. This work aims to rectify this shortcoming through detailed evaluation of prevalent TCP variants in dierent topology settings over the AODV routing protocol. Subsequent results reveal the performance merits of TCP Vegas and NewReno in MANETs with respect to Reno which is further explored and accounted for. Finally, insight is provided through extensive

Early in the course of MANET research, TCP behaviour came into focus as it became understood that TCPs misreaction to noncongestion induced packet losses results in suboptimal performance in such wireless environments. Packet losses in wired networks are largely attributed to buer contention at branching points in the network topology, but in mobile ad hoc networks losses due to errors or mobility may be at least as frequent. The appropriate reaction to each of these causes is dierent to that of contention losses but as TCP is unable to distinguish between them, it misreacts. Further research has revealed that the widely implemented 802.11 distributed MAC mechanism can severely penalise TCP ows when competing with other congestion aware or unaware trac. This may lead to a capture eect whereas some TCP ow (or ows) severely under-utilise the oered bandwidth. Several studies have examined the behaviour of TCP variants under specic conditions to highlight a particular performance aspect and suggest modications. However, there have been few detailed comparison studies among dierent TCP variants. Earlier research work by Ramarathinam et al. [16] examined the goodput performance of TCP Tahoe, Reno, NewReno and SACK in a static multi-hop network under three dierent routing protocols and noted that overall, Reno is the

50

6.3.2

AODV INTERACTION WITH in subsequent sections. The scenario topolTCP ogy consists of a simple ve node string topology (nodes A ! E and an additional This section examines the behaviour of TCP node (node F) which is initially disconnected Reno in asimple mobile ad-hoc network sce- from the network, as shown in Figure 1(a). nario. This example is instructive on the The nodes in the string are spaced 200m challenges faced by TCP in a wireless multi- apart and feature standard LucentWaveLan hop environment and provides insight on the transceivers with bandwidth of 2Mbps. The interaction of TCP with the AODV routing transmission range of the transceivers is xed mechanism. In particular, the discussion that at 250m using a at, ideal signal propagation follows aids comprehension when contemplat- model under which there is no attenuation up ing throughput performance issues with TCP to the transmission range limit and nullies in subsequent sections. The scenario topol- the signal strength beyond that limit. ogy consists of a simple ve node string topology (nodes A ! E and an additional node (node F) which is initially disconnected from 6.3.4 SIMULATION RESULTS the network, as shown in Figure 1(a). The The goodput results for the TCP agents unnodes in the string are spaced 200m apart der examination, namely Reno, NewReno, and feature standard LucentWaveLan II [11] and Vegas, in strip topologies are contained transceivers with bandwidth of 2Mbps. The in Figure IV as a function of goodput over transmission range of the transceivers is xed maximum node speed under continuous moat 250m using a at, ideal signal propagation bility conditions. The error bars represent 90 model under which there is no attenuation up condence intervals. In these experiments, to the transmission range limit and nullies TCP Vegas seems to performs signicantly the signal strength beyond that limit. better than NewReno; however, the dier6.3.3 AODV INTERACTION WITH TCP ence could potentially diminish if the agent is adversely aected by other factors, such as the interaction with other trac in the network. Nonetheless, the results presented here conrm Vegas competent performance (on par with NewReno) under various mobility conditions when interaction with other trafc is not taken into account. This expands on previous work comparing Vegas with Reno [13] and reveals that Vegas performance merits are equivalent or more pronounced than NewRenos when single connections are considered. In our experiments Vegas had con-

This section examines the behaviour of TCP Reno in asimple mobile ad-hoc network scenario. This example is instructive on the challenges faced by TCP in a wireless multihop environment and provides insight on the interaction of TCP with the AODV routing mechanism. In particular, the discussion that follows aids comprehension when contemplating throughput performance issues with TCP

51

sistently higher goodput than NewReno (2- 6.4 4

6.3.5

CONCLUSION

Multimedia Applications for MANETs over Homogeneous and Heterogeneous Mobile Devices

This research has examined the performance of three TCP variants over dynamic topologies in MANETs. The discussion included detailed traces of a route breakage scenario, studied the eect on TCP throughput and provided insight on the interplay of route buering and the transport agent. Further, TCP Reno, NewReno and Vegas were evaluated in dynamic topologies over square and strip simulation areas. The performance merit of TCP Vegas in terms of throughput was noted as well as the almost equivalent eectiveness of TCP NewReno. Reno was shown conclusively to be the worst performer under all mobility conditions. An explanation of discrepancy of the TCP agents performance over a stable path was provided. Future research prospects include the quantication of the causes of packet loss and the examination of aggregate multiple ow behaviour. It may be of particular interest to explore to what degree hidden terminal effects, or other MAC layer incompatibilities are responsible for the degradation in TCP throughput, compared to the decline caused by route breakages. Finally, there is significant value in examining the behaviour of multiple TCP ows for dierent TCP agents; however, such an examination should be accompanied by some metric of interaction between the ows in order for the results to be meaningful.

Mobile Ad Hoc Networks (MANETs) are considered a vital part in beyond third generation wireless networks (Nicopolitidis et al., 2003). In the matter of fact, they present a new wireless networking paradigm. Any sort of xed infrastructure is not used byMANETs. They are important sorts of WLANs, therefore, in a distributed and a cooperative environment, MANETs do eciently function (Murthy and Mano, 2004) (Sarkar et al., 2008). MANETs are networks of self-creating since there is a lack of routers, conguration prior to the network setup, Access Points (APs) and predetermined topology (Wu et al., 2007). MANETs are as well networks of self-administering and selforganizing. This is because in the network creation process, there is no application for central control. On MANETs, it is extremely hard to apply any of the central administration types, for instance, congestion control due to the dynamic nature of the network topology inMANETs, authentication or central routing. In short, several important applications beneted from MANETs, for example, in military, ubiquitous, emergency and collaboration computing. 6.4.1 Introduction

In this chapter, describe the necessary background for the MANETs over homogeneous and heterogeneous mobile devices. The re-

52

searcher begin this chapter to introduce the related background and main concepts of the Mobile Ad Hoc Network (MANETs) in Section, and explained briey about the existing wireless mobile network approaches, wireless ad hoc networks, wireless mobile approaches in Section The characteristic of MANETs are in Section The types of Mobile Ad hoc network The trac types in ad hoc networks which include the Infrastructure wireless LAN and ad hoc wireless LAN are presented highlight the relevant details about the ad hoc network routing protocol performance issues. The types of ad hoc protocols such as (Table-driven, On-demand and Hybrid) and Compare between Proactive versus Reactive and Clustering versus Hierarchical respectively. The existing ad hoc protocols are presented in Section. The four important issues signicant in MANET are Mobility, QoS Provisioning, Multicasting and Security is presented in Section Furthermore, the practical application and the MANET layers are shown in Section and Section 1.2.12 respectively . Finally, in Section the summary of this chapter. 6.4.2 ISSUES IN DESIGNING TRANSPORT LAYER A

In this section, some of the issues to be considered while designing a transport layer protocol for ad hoc wireless networks are discussed. I. Induced trac: Unlike wired networks, ad hoc wireless networks utilize multi-hop radio relaying. A link-level transmission aects the neighbor nodes of both the sender and re-

ceiver of the link. In a path having multiple links, transmission at a particular link aects one upstream link and one downstream link. This trac at any given link (or path) due to the trac through neighboring links (or paths) is referred to as induced trac. This is due to the broadcast nature of the channel and the location-dependent contention on the channel. This induced trac aects the throughput achieved by the transport layer protocol. II. Induced throughput unfairness: This refers to the throughput unfairness at the transport layer due to the throughput/delay unfairness existing at the lower layers such as the network and MAC layers. For example, an ad hoc wireless network that uses IEEE 802.11 DCF as the MAC protocol may experience throughput unfairness at the transport layer as well. A transport layer protocol should consider these in order to provide a fair share of throughput across contending ows. III. Separation of congestion control, reliability, and ow control: A transport layer protocol can provide better performance if end-to-end reliability, ow control, and congestion control are handled separately. Reliability and ow control are end-to-end activities, whereas congestion can at times be a local activity. The transport layer ow can experience congestion with just one intermediate link under congestion. Hence, in networks such as ad hoc wireless networks, the performance of the transport layer may be improved if these are separately handled. While separating these, the most important objective to be considered is the minimization of

53

the additional control overhead generated by them. IV. Power and bandwidth constraints: Nodes in ad hoc wireless networks face resource constraints including the two most important resources: (i) power source and (ii) bandwidth. The performance of a transport layer protocol is signicantly aected by these constraints. V. Misinterpretation of congestion: Traditional mechanisms of detecting congestion in networks, such as packet loss and retransmission timeout, are not suitable for detecting the network congestion in ad hoc wireless networks This is because the high error rates of wireless channel, location-dependent contention, hidden terminal problem, packet collisions in the network, path breaks due to the mobility of nodes, and node failure due to a drained battery can also lead to packet loss in ad hoc wireless networks.

wireless networks are not addressed. This causes reduction of throughput when TCP is used in ad hoc wireless networks. It is very important to employ TCP in ad hoc wireless networks as it is important to seamlessly communicate with the Internet whenever and wherever it is available. This chapter provided a discussion on the major reasons for the degraded performance of traditional TCP in ad hoc wireless networks and explained a number of recently proposed solutions to improve TCPs performance. Other non-TCP solutions were also discussed in detail. The second half of this chapter dealt with the security aspect of communication in ad hoc wireless networks. The issues and challenges involved in provisioning security in ad hoc wireless networks were identied. This was followed by a layer-wise classication of the various types of attacks. Detailed discussions on key management techniques and secure routing techniques for ad hoc wireless networks were provided. Table 9.4 lists out the 6.4.3 SUMMARY various attacks possible in ad hoc wireless This chapter discussed the major challenges networks along with the solutions proposed that a transport layer protocol faces in ad for countering those attacks. hoc wireless networks. The major design goals of a transport layer protocol were listed and a classication of existing transport layer solutions was provided. TCP is the most widely used transport layer protocol and is considered to be the backbone of todays Internet. It provides end-to-end, reliable, byte-streamed, in-order delivery of packets to nodes. Since TCP was designed to handle problems present in traditional wired networks, many of the issues that are present in dynamic topology networks such as ad hoc 54

Image0911.jpg

Figure 1: A Sample diagram

55