You are on page 1of 51

A Joint Approach To Cross Layer

Routing Metric And Cost Based


Rate Adaptation
Department of Information Science & Engineering,
BVBCET Hubli
May 26, 2014
ACKNOWLEDGMENT
The sense of contentment and elation that accompanies the successful
completion of the rst phase of our project and its report would be in-
complete without mentioning the names of the people who helped us in
accomplishing this.
We take this opportunity to thank our principal Dr. Ashok Shettar, for
providing healthy environment in the college, which helped in concentrating
on the task. We express a deep sense of gratitude to our H. O. D. Prof.
Meena. S. M. for providing the inspiration required for taking the project
to its completion.
We sincerely thank Prof. Narayan D. G. for his inspiring guidance and
promising support they gave during the course of the project.
Last but not the least we like to thank all the sta members, teaching
and non - teaching sta for helping us during the course of the project.
Mr. Anup Shanbhag
Mr. Rahul Hosmani
Ms. Sana Chandsha
Mr. Vinayak Anegundi
ABSTRACT
A wireless mesh network (WMN) is a communications network made up
of radio nodes organized in a mesh topology. Wireless mesh networks often
consist of mesh clients, mesh routers and gateways. The mesh clients are
often laptops, cell phones and other wireless devices while the mesh routers
forward trac to and from the gateways which may, but need not, connect
to the Internet. The coverage area of the radio nodes working as a single
network is sometimes called a mesh cloud. Access to this mesh cloud is
dependent on the radio nodes working in harmony with each other to create
a radio network. A mesh network is reliable and oers redundancy. When
one node can no longer operate, the rest of the nodes can still communicate
with each other, directly or through one or more intermediate nodes.Mesh
networks may involve either xed or mobile devices. The solutions are as
diverse as communication needs, for example in dicult environments such
as emergency situations, tunnels, oil rigs, battleeld surveillance, high speed
mobile video applications on board public transport or real time racing car
telemetry. An important possible application for wireless mesh networks is
VoIP. By using a Quality of Service scheme, the wireless mesh may support
local telephone calls to be routed through the mesh. Here we developed a
new Routing Metric which considers link load, loss ratio and interference.
And then applied a Rate adaptive mechanism which further improved the
performance of the metric.
Keywords :MultiRadio, Rate adaption, wireless mesh networks
i
Contents
1 Introduction 2
1.1 Literature Survey: . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 Problem Statement . . . . . . . . . . . . . . . . . . . . . . . 10
1.3 Scope of the the project . . . . . . . . . . . . . . . . . . . . . 11
2 REQUIREMENT ANALYSIS 12
2.1 System Model . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.2 Functional Requirements . . . . . . . . . . . . . . . . . . . . . 12
2.2.1 Module 1: Incorporation of multiple interface for OLSR. 12
2.2.2 Module 2: Proposal of new Routing metric . . . . . . 13
2.2.3 Module 3: Incorporation of Rate Adaptive Transmis-
sion in current NS-2 . . . . . . . . . . . . . . . . . . . 13
2.3 Software and Hardware requirement . . . . . . . . . . . . . . 14
2.3.1 Software Requirements . . . . . . . . . . . . . . . . . . 14
2.3.2 Hardware Requirements . . . . . . . . . . . . . . . . . 14
3 SYSTEM DESIGN 15
3.1 Architecture Design . . . . . . . . . . . . . . . . . . . . . . . 15
3.2 Data Design . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.2.1 Packet Format for OLSR protocol . . . . . . . . . . . 16
3.2.2 Packet format for Hello message . . . . . . . . . . . . 18
3.2.3 Packet format for TC message . . . . . . . . . . . . . 19
3.3 Component Level Design . . . . . . . . . . . . . . . . . . . . . 20
3.3.1 Link Sensing . . . . . . . . . . . . . . . . . . . . . . . 20
3.3.2 Neighbor Detection . . . . . . . . . . . . . . . . . . . . 21
3.3.3 Multipoint Relay Selection . . . . . . . . . . . . . . . 22
3.3.4 Topology Control Message Diusion . . . . . . . . . . 23
3.3.5 Routing Table calculation . . . . . . . . . . . . . . . . 24
3.3.6 Rate Adaptation Algorithm . . . . . . . . . . . . . . . 24
ii
4 IMPLEMENTATION 26
4.1 New Routing Metric CLIA . . . . . . . . . . . . . . . . . . . . 26
4.1.1 Implementation in OLSR pkt.h and repositories.h . . 27
4.1.2 Implementation in mac-timers.h . . . . . . . . . . . . 29
4.1.3 Implementation in mac-802 11.cc . . . . . . . . . . . 30
4.1.4 Implementation in OLSR.cc . . . . . . . . . . . . . . . 33
4.1.5 Implementation in cbr trac.cc . . . . . . . . . . . . 33
5 Results and Discussion 35
5.1 Performance metrics: . . . . . . . . . . . . . . . . . . . . . . . 35
5.2 Simulation Environment . . . . . . . . . . . . . . . . . . . . . 36
6 Conclusion and Future Work. 43
6.1 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
6.2 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
iii
List of Figures
1.1 Wireless Mesh Network Architecture . . . . . . . . . . . . . . 3
1.2 WCETT-LB . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3 CATT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.4 ILA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.5 CWB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.6 MIND . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.7 C2WB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.8 iAWARE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.9 Link Sensing . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.10 Neighbor Detection . . . . . . . . . . . . . . . . . . . . . . . . 10
2.1 System model . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
3.1 System Architecture . . . . . . . . . . . . . . . . . . . . . . . 16
3.2 OLSR packet format . . . . . . . . . . . . . . . . . . . . . . . 17
3.3 Hello message packet format . . . . . . . . . . . . . . . . . . . 18
3.4 Hello message packet format . . . . . . . . . . . . . . . . . . . 20
4.1 CLIA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
5.1 Scenario I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
5.2 Throughput Vs Packet Rate of Scenario I . . . . . . . . . . . 37
5.3 End to end delay Vs Packet Rate of Scenario I . . . . . . . . 37
5.4 PDF Vs Packet Rate of Scenario I . . . . . . . . . . . . . . . 38
5.5 Scenario II . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
5.6 Throughput Vs Packet Rate of Scenario II . . . . . . . . . . . 39
5.7 End to end delay Vs Packet Rate of Scenario II . . . . . . . . 39
5.8 PDF Vs Packet Rate of Scenario II . . . . . . . . . . . . . . . 40
5.9 Scenario III . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
5.10 Throughput Vs Packet Rate: CLIA vs CLIA-RA . . . . . . . 41
5.11 Scenario IV . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
iv
Wireless Mesh Networks
5.12 Throughput Vs Packet Rate: CLIA vs CLIA-RA . . . . . . . 42
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 1
Wireless Mesh Networks
Chapter 1
Introduction
A wireless mesh network is a communications network made up of radio
nodes organized in a mesh topology. Wireless mesh networks often consist
of mesh clients, mesh routers and gateways. The mesh clients are often lap-
tops, cell phones and other wireless devices while the mesh routers forward
trac to and from the gateways which may, but need not, connect to the
Internet. Wireless mesh networks hold great promise in the wireless trans-
mission, particularly if the problem of providing sucient network capacity
can be addressed. As mesh nodes are equipped with multiple radios, they
introduce intra-ow and interow interference. Thus, there is a need for new
routing metrics and techniques to improve Quality of Service. Many cross
layer routing metric are proposed in the literature survey. We propose a new
metric by considering the drawbacks of two well-known metrics Interference
Aware routing (iAWARE) and Estimated Transmission Count (ETX). The
new routing metric will help us chose the optimum path by considering
the various parameters like interference, delay. Later, we address the joint
problem of routing and application layer rate adaption using routing metric
parameters.
Wireless Mesh Networks can be classied into three types as follows:
Infrastructure wireless mesh network: An infrastructure wire-
less mesh network (WMN) is a hierarchical network consisting of mesh
clients, mesh routers and gateways. Mesh routers constitute a wireless
mesh backbone, to which mesh clients are connected as a star topol-
ogy, and gateways are chosen among mesh routers providing Internet
access.
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 2
Wireless Mesh Networks
Client WMN: Client WMN provides peer-to-peer networking facil-
ity between the client nodes. Client nodes are normally single radio
devices with optional routing capability depending on the end user
requirements. Client WMNs are very simple to design but scalability
and limited resource allocation are the critical issues which may cause
throughput degradation in the network.
Hybrid WMN:Hybrid WMNs consist of both hierarchical and client
architectures. Mesh routers are equipped with the bridging function-
ality in order to integrate with other networks like wired networks,
WiFi, MANETs and VANETs is an example of hybrid WMNs.
Figure 1.1: Wireless Mesh Network Architecture
1.1 Literature Survey:
ETX[1] is dened as the number of expected transmission plus re-
transmissions required to successfully deliver a packet over a wireless
link. ETX is an additive metric.Expected transmission count and re-
transmissions required to successfully deliver a packet over that link.
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 3
Wireless Mesh Networks
ETX =1/df.dr
df:packet delivery ratio in forward direction
dr:packet delivery ratio in back ward direction
Poor performance in ETX is due to assumption that the channel condi-
tions are static in nature. It does not measure trac load and interference
on the network.
ETT[2]Expected transmission time; It measures the transmission rate
of individual link.
ETT = ETX x (S / B)
S:Average packet size
B:Data rate
ETT is an improved version of ETX, but still has drawbacks of being
unaware of trac load, intra-ow interference and inter-ow interference.
WCETT-LB[3]Extension of ETT which considers packet loss ratio
and transmission rate of each link. It also provides load balancing to avoid
congestion.
Figure 1.2: WCETT-LB
L(path): load balancing component
QLi: mean/average queue length
Bi: transmission rate which captures level of congestion.
Ni: total number of child nodes using node i as their next hop path on p.
CATT[4]Contention aware transmission time It predicts location de-
pendent Contention and rate diversity in the links. Isotonic behavior of
CATT makes it possible to work with link state routing protocols.
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 4
Wireless Mesh Networks
Figure 1.3: CATT
Ni, Nj:Total number of links interfering in transmission on link I and
link j respectively.
Rk, Rj: Packet size of link containing 1 and 2 hop neighbors respectively.
Bk, Bj: Bandwidth of links 1 and 2 hop neighbors respectively.
Tj: packet transmission attempt rate on link j.
It captures interow and intraow interference simultaneously. But it as-
sumes that all the neighboring nodes to participate in the interference
(which is an over estimate).It also uses active probing which has an over-
head.
ILA[5]Interference load aware routing metric It is a hybrid metric
based on load, interfering neighbors and transmission rates.It measures
inter ow, intra ow interferences, transmission rates and packet loss ra-
tio.
Figure 1.4: ILA
P: path in the network.
MTI: Metric of Trac interference.
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 5
Wireless Mesh Networks
CSC: (Channel Switching Cost) measures eciency of ows routed through
path p.
MTI measures quantity of trac generated by interfering neighboring
nodes.
ETT:It measures the dierence in transmission rate and packet loss ratio.
AIL: Average interfering Load (when node i and node j are transmitting
over the channel C)
Nl(C): set of interfering neighbor of node i and j
ILij: interference load of neighbors.
CSC: Channel Switching Cost, it captures intra ow interference
: It is a scaling factor to balance the eect of MTI and CSC
min(ETT): smallest ETT
min(AIL): average load ILA
ILA uses active probing (overhead high) which does not consider trans-
mission delay in order to route the trac eciently.
CWB[6] Contention Window based Routing metric.It consider chan-
nel utilization and average contention window.
It is a load interference routing metric which guides the routing protocol
to balance trac loads on the links and increase throughput by routing to-
wards less congested areas.
Figure 1.5: CWB
FER: Frame Error Rate
CW0: Minimum contention window
r: maximum back o stage
: (Channel utilization component )Channel Busy time CBT
CWB captures inter ow and trac load but fail to capture intra ow
interference. This metric is poor when network changes quickly.
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 6
Wireless Mesh Networks
MIND[7]Metric for interference and channel diversity.It captures in-
terference and load aware parameters on the basis of passive monitoring
technique.
MIND captures both inter ow and intra ow interference in a intelligent
manner by considering both physical and logical models. It induces com-
plexity in implementation in virtual Networks.
Figure 1.6: MIND
INTERLOAD:captures interow interference and load
CSC:captures intra ow interference
n: measures total number of wireless links
m: total number of nodes in the path p.
IR:Interference Ratio, it is used to measure interference between links
SINR:Signal to noise plus interference ratio
SNR:Signal to noise ratio
Total time: It is the measure of time form rst attempt to the receiving
of ACK.
Idle Time: Back o time in radio nodes.
MIND captures both inter ow and intra ow interference in a intelligent
manner by considering both physical and logical models. It induces com-
plexity in implementation in virtual Networks.
C2WB[8]C2WB is a routing metric that leverages real-time radio
channel quality information to maximize the achieved throughput of a
WMN. It uses service time as a metric to approximate quality of the link
without using any heuristic approaches.
CWl: Average Contention window of the node link l
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 7
Wireless Mesh Networks
Figure 1.7: C2WB
FER: Frame Error Rate.
Tbac, l: Back o time of the link
Idle Time: Back o time in radio nodes.
C2WB measures intra ow interference, data rate and frame error rate to
compute the most optimal path for transmission.
iAWARE[9]iAWARE stands for Interference Aware.
This routing metric addresses the problem of interference by combining
the interference ratio (IR) with the ETT metric.
Figure 1.8: iAWARE
iAWARE has no MAC layer interference measurement mechanism, as it
only capture the interference at a node level. Lack of load balancing pa-
rameters may lead the trac to a congested route.
OLSR[10]Optimized Link State Routing Protocol is a proactive link
state routing protocol, which uses hello and topology control messages
to discover and then disseminate link state information throughout the
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 8
Wireless Mesh Networks
mobile ad-hoc network. The functionality of OLSR species the behavior
of a node, equipped with OLSR interfaces participating in the MANET
and running OLSR as routing protocol.
Link sensing: Link sensing is done to detect the type of link exist
between local interface and the remote interface. Link Sensing is ac-
complished through periodic emission of HELLO messages over the
interfaces through which connectivity is checked. A separate HELLO
message is generated for each interface. Resulting from the link sens-
ing is the local link set, describing links between local interfaces and
interfaces.
Figure 1.9: Link Sensing
In the gure 1.9, nodes A and B runs OLSR on multiple interfaces. B
uses the address of b1 as its main address. Nodes D and C runs on
single interfaces (d1 and c1). To keep up-to-date information on what
links exist between a node and its neighbors, the link set is maintained.
Neighbor Detection: Neighbor detection is done by a node to detect
the type of its neighbor. Neighbor detection populates the 1-hop neigh-
bor repository and only uses the main addresses of nodes. Whenever a
link entry is created, the neighbor table is queried for a corresponding
neighbor entry. A neighbor is said to be a symmetric neighbor if there
exists at least one Link-entry in the link set connecting a local inter-
face to one of the neighbors interfaces where the symmetric timer is
not timed out. When a link-entry is deleted, the corresponding neigh-
bor entry is also removed if no other link entries exist for this neighbor.
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 9
Wireless Mesh Networks
Figure 1.10: Neighbor Detection
The Figure 1.10 represents the neighbor detection process. This is
done through the exchange of HELLO message packets. If node B
wants to nd its neighbor it oods the HELLO message through the
whole topology. The nodes which are its neighbors will signal the node
B by sending the HELLO message packet back to it. In this case the
nodes A, C, D are detected as its neighbors.
Multipoint Relay Selection: The concept of multipoint relaying is
to reduce the number of duplicate retransmissions while forwarding a
broadcast packet. This technique restricts the set of nodes retransmit-
ting a packet from all nodes, to a subset of all nodes. Each node selects
its MPR set from among its 1-hop symmetric neighbors. This set is
selected such that it covers all symmetric strict 2-hop nodes. The in-
formation required to perform this calculation is acquired through the
periodic exchange of HELLO messages. The MPR set is re-calculated
when a change in 1-hop or 2-hop Neighbor sets with bi-directional link
is detected.
The MPR selection process reduces the number of retransmissions.
The gure at the left side shows the process of ooding without MPR
selections. So the number of retransmissions increases. If the ooding
is done by only considering the MPRs the retransmission decreases
which reduces the trac.
1.2 Problem Statement
To propose a cross layer rate adaptive technique for data transmis-
sion in multi-radio wireless mesh networks based on a new routing metric
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 10
Wireless Mesh Networks
and adapt the transmission rate at application layer based on routing
metric parameters.
1.3 Scope of the the project
To integrate the existing OLSR routing protocol with Multiple Inter-
faces in NS-2.
To design and implement a new Routing metric for multiple radio
wireless mesh networks.
To implement rate adaptive mechanism to improve the quality of ser-
vice.
To carry out extensive performance analysis of proposed techniques
using various random topologies and trac.
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 11
Wireless Mesh Networks
Chapter 2
REQUIREMENT
ANALYSIS
This chapter deals with the requirements of the entire system.It also spec-
ies the methodologies to achieve the requirements specied. In this
chapter we specify about the System model, Functional Requirements,
Performance requirements, Software Quality Attributes and the project
plan
2.1 System Model
The system model as shown in gure 2.1 represents the cross layer
routing for wireless mesh network which has parameter acquisition from
dierent layers and using them in a routing metric to select an optimum
path for transmission in the wireless network.
2.2 Functional Requirements
This part deals with the various functional requirements of the sys-
tem.
2.2.1 Module 1: Incorporation of multiple interface for
OLSR.
The existing ns-2 OLSR does not have multiple interfaces.
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 12
Wireless Mesh Networks
Figure 2.1: System model
The task is to scale up the existing OLSR protocol and incorporate
multiple interfaces.
2.2.2 Module 2: Proposal of new Routing metric
Choosing the optimal path in wireless networks very tedious. There
are many metrics proposed for calculation of optimal path. We propose
a new metric by considering the drawbacks of two well-known metrics
Interference Aware routing (iAWARE) and Estimated Transmission
Count (ETX).
The task is to propose a new routing metric which helps in calculating
the optimal path in the network for data transmission.
It considers link load, loss ratio and interference.
2.2.3 Module 3: Incorporation of Rate Adaptive Trans-
mission in current NS-2
Rate adaption is a technique of reducing the rate of transmission in
the network whenever congestion occurs.
The existing NS-2 does not have this facility. The task is to incorporate
Rate Adapative transmission in existing NS-2 to further improve the
performance.
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 13
Wireless Mesh Networks
2.3 Software and Hardware requirement
There are many additional hardware and software which are re-
quired for the development of any software application; they also consti-
tute a major part of the application. This section provides the overview
of the additional requirements used in the development of this application.
2.3.1 Software Requirements
Implementation of the routing metric and the analysis will be done
with C++ programming as backend coding.
Operating System: Linux Open SUSE.
Topology creation will be done using TCL scripts as front-end coding
and the OLSR protocol changes was done using C++ programming.
2.3.2 Hardware Requirements
The system should contain minimum of 2GB RAM and 256MB GPU.
Since the project is completely based on simulation no explicit hard-
ware is required.
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 14
Wireless Mesh Networks
Chapter 3
SYSTEM DESIGN
A software design document (SDD) is a written description of a software
product, that gives a software development team an overall guidance of
the architecture of the software project. An SDD usually accompanies
an architecture diagram with pointers to detailed feature specications of
smaller pieces of the design. Practically, a design document is required to
coordinate a large team under a single vision. A design document needs to
be a stable reference, outlining all parts of the software and how they will
work. The document is commanded to give a fairly complete description,
while maintaining a high-level view of the software.
3.1 Architecture Design
System design phase is a main core part of software development.
System design takes software requirement specications as input and it
concerns on the system development by providing a call return architec-
ture and algorithm for the modules, So that the coding of the system
will be easily carried out. It also decides the interconnection between the
modules.
The gure shows the system architecture of the project. Link sensing
updates the link set of local information base. The HELLO message is
used for link sensing. The processing of various packets is done by packet
processing module. Then routing metric measurements are made on the
link to select the ecient route. Whenever a link tuple is created neighbor
detection is carried out by each node which updates the 1-hop and 2-hop
neighbor set. For the ecient ooding and to reduce the trac which
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 15
Wireless Mesh Networks
happens through the ooding MPR are selected from this neighbor set so
that there is no duplication created. This is done by MPR selection. This
discovered topology information is diused in the network through the TC
message using the default forwarding algorithm. Routing table calculation
module takes the information from these all modules and calculate the
routing table hence the ecient route to route the packet.
Figure 3.1: System Architecture
3.2 Data Design
This section involves dierent message or packet formats and the
repositories.
3.2.1 Packet Format for OLSR protocol
The basic layout of any packet in OLSR is as follows (omitting IP and
UDP headers):
Packet Length: The length in bytes of the entire packet, including
the header
Packet Sequence Number: A sequence number incremented by one
each time a new OLSR message is transmitted by this host. A sepa-
rate Packet Sequence Number is maintained for each interface so that
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 16
Wireless Mesh Networks
Figure 3.2: OLSR packet format
packets transmitted over an interface are sequentially enumerated.
Message type:An integer identifying the type of this message. Mes-
sage types of 0-127 are reserved by OLSR while the 128-255 space
is considered private and can be used for custom extensions of the
protocol.
Vtime: This eld indicates for how long after reception a node will
consider the information contained in the message as valid. The time
interval is represented in a mantissa-exponent format.
Message Size:The size of this message, including message header,
counted in bytes.
Originator Address:Main address of the originator of this message.
Time To Live: The maximum number of hops this message can be
forwarded. Using this eld one can control the radius of ooding.
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 17
Wireless Mesh Networks
Hop Count: The number of times the message has been forwarded.
Message Sequence Number: A sequence number incremented by
one each time a new OLSR packet is transmitted by this host.
3.2.2 Packet format for Hello message
Figure 3.3: Hello message packet format
Reserved: This eld must be set to 0000000000000 to be in com-
pliance with this specication.
HTime: This eld species the HELLO emission interval used by the
node on this particular interface
Willingness: Willingness is set for any of the following values.
1. WILL NEVER = 0
2. WILL LOW = 1
3. WILL DEFAULT = 3
4. WILL HIGH = 6
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 18
Wireless Mesh Networks
5. WILL ALWAYS = 7
Link code structure: Link Code value holds two dierent elds, of
two bits each they are Neighbor Type and Link Type.
1. Link types:
(a) UNSPEC LINK - indicating that no specic information
about the links is given.
(b) ASYM LINK - indicating that the links are asymmetric.
(c) SYM LINK - indicating that the links are symmetric with
the interface.
(d) LOST LINK - indicating that the links have been lost.
2. Neighbor Types:
(a) SYM NEIGH - indicating that the neighbors have at least
one symmetrical link with this node.
(b) MPR NEIGH - indicating that the neighbors have at least
one symmetrical link AND have been selected as MPR by
the sender.
(c) NOT NEIGH - indicating that the nodes are either no longer
or have not yet become symmetric neighbors.
Link Message Size: The size of the link message, counted in bytes
and measured from the beginning of the Link Code eld and until
the next Link Code eld.
Neighbor Interface Address: The address of an interface of a
neighbor node.
3.2.3 Packet format for TC message
Advertised Neighbor Sequence Number (ANSN): Sequence
number is associated with the advertised neighbor set. Every time
a node detects a change in its advertised neighbor set, it increments
this sequence number. This number is sent in this ANSN eld of the
TC message to keep track of the most recent information.
Multipoint Relay Selector address: This eld contains the main
address of the node which selected the MPR.
QoS elds values: This eld represents the various QoS parameters
which need to be considered.
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 19
Wireless Mesh Networks
Figure 3.4: Hello message packet format
3.3 Component Level Design
Component-level design establishes the algorithmic detail required to ma-
nipulate data structures, eect communication between software compo-
nents via their interfaces, and implement the processing algorithms allo-
cated to each component. The below design describes the OLSR protocol.
3.3.1 Link Sensing
Link Sensing is accomplished through periodic emission of HELLO mes-
sages. This is done for detection of local link set i.e. all the neighboring
nodes are sensed and a neighboring nodes set is calculated.
Linksensing()
ALGORITHM linksensing()
// Performs Hello Message generation and updates its Link set
// Input: none
// Output: Updated Linkset
While true
generateHelloMessage()
OP <- Read HelloMessage from Neighbour Node
processHelloMessage(OP)
Delay 2 sec
processHelloMessage(OP)
ALGORITHM processHelloMessage()
// Processes the HelloMessage and updates the link set
// Input: OLSR Packet with HelloMessage
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 20
Wireless Mesh Networks
// Output: Updates Link Set
If OP->vtime >current time
If not SourceAddr= L InterfaceAddr in LinkSet
Create new tuple and update
L Neigbour iface addr=sourceaddr
L local iface addr = Address of the Interface received
hello message
L SYM Time = current time 1 (expired)
L time = current time + validity time
Else //tuple exists
Modify tuple as
L ASYM time=current time + validity time
if OP->iaddr = OP->Message->iAddr
if L linktype = LOST LINK
L SYM time=current time 1
else if L linktype is SYM LINK or ASM LINK
L SYS time=current time + validity time
L time = L SYM time + NEIGB HOLD TIME
L time=max(L time, L ASYM time)
3.3.2 Neighbor Detection
First all the routes to destination are found. Then all the neighbors and
the links are detected and used for transmission of packets.
processHelloMessage(OP)
ALGORITHM processHelloMessage()
// Processes the HelloMessage and updates the link set
// Input: OLSR Packet with HelloMessage
// Output: Updates Link Set
If OP ->vtime >current time
If not SourceAddr=L InterfaceAddr in LinkSet
Create New tuple and update
L Neigbour iface addr=sourceaddr
L local iface addr = Address of the Interface received
hello message
L SYM Time = current time 1 (expired)
L time = current time + validity time
Else //tuple exists
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 21
Wireless Mesh Networks
Modify tuple as
L ASYM time=current time + validity time
if OP->iaddr = OP->Message->iAddr
if L linktype = LOST LINK
L SYM time=current time 1
else if L linktype is SYM LINK or ASM LINK
L SYS time=current time + validity time
L time = L SYM time + NEIGB HOLD TIME
L time=max(L time, L ASYM time)
UpdateNeigborSet()
ALGORITHM updateNeigbourSet()
// Triggers when there is some updation in Link Set
//Input: Changes in Link Tuple
// Output: Updates N status in neighbor set
If L SYM time <= current time
set N status to SYM
Else
set N status to NOT SYM
CreateNeigborSet()
ALGORITHM createNeigbourSet()
// Triggers when new Link tuple is Deleted
// Input: Changes in Link Tuple
// Output: Removes tuple from neighbor set
MainAddr <- getRemovedAddrfromlinkset (L neighbour iface addr)
// Fetches main address from the Link set
Delete Neighbor Set tuple with N neighbor main addr as MainAddr
3.3.3 Multipoint Relay Selection
Each node selects its MPR set independently. This set contains subset
of 1-hop neighbors which provide maximum reachability guarantees from
2-hop neighbors to given node.
ALGORITHM: MPR Selection
//Selects MPRs from the set of 1 hop neighbor which covers maximum
set of 2-hop neigbor set
//Input: selected 1-hop neighbor and 2-hop neighbor set
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 22
Wireless Mesh Networks
//Output: MPR set
Start with an empty MPR set MPR(x);
While there still exist nodes N(x)
Let y be one of the node in N(x)
y be the only node from N(x) provides reachability to a node in
N2(x)
add y to MPR set
In case of a tie in the above step, select that node which is
already in MPR(x).
Mark y as covered
End
3.3.4 Topology Control Message Diusion
Topology control messages are broadcast and retransmitted by the MPRs
with the purpose of providing each node in the network with sucient
link state information to allow route calculation.
ALGORITM TC msg diffusion
// Provides each node in the network, link state information to
allow route calculation.
//Input:OLSR packet received,message to be sent, interface address.
//Output:none
Compute validity time from Vtime field
If sender iface addr=iface addr and ttl >1
Retransmit the message
If there exist some tuple in the topology sset where:
T last addr == originator address AND
T seq >ANSN
Discard the message If T last addr == originator address AND
T seq <ANSN
Remove the tuple from the topology set
For each of the advertised neighbor main address received in the
TC message:
If there exist some tuple in the topology set where:
T dest addr == advertised neighbor main address, AND
T last addr == originator address,
then the holding time of that tuple MUST be set to:
T time = current time + validity time.
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 23
Wireless Mesh Networks
else
T dest addr = advertised neighbor main address,
T last addr = originator address,
T seq = ANSN,
T time = current time + validity time.
3.3.5 Routing Table calculation
Routing table calculation is done by taking the information from TC mes-
sage which declares the topology in the network.
Algorithm: Routing table calculation
//triggers when there is some changes in linkset, neighbor set,
MPR, TC message set, multiple information base set.
//input: changes in repositories, r table
//output: routing table(gives the appropriate route to destination)
if changes in link set
for each changes in link set
set r table field (i) <- link tuple(i)
if changes in neighbor set
for each changes in 1 hop neighbor set
set r table field(i) <- 1 hop neighbor tuple(i)
if changes in 1 hop neighbor set
for each changes in 2 hop neighbor set
set r table fields(i) <- 2 hop neighbor tuple
if changes in topology cont msg set
for each changes in topology cont msg set
set r table fields(i) <- topology cont msg tuple
if changes in multi asso base set
for each changes in multi asso base set(i)
set r table fields(i) <- multi asso base tuple(i) update r
table
3.3.6 Rate Adaptation Algorithm
ALGORITHM RateAdaptation
//Input: ETX(Expected Transmission Count), IR(Interference Ratio),
TxRate(Initial Transmission Rate), LB(Lower Bound), UB(Upper Bound)
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 24
Wireless Mesh Networks
//Output: New TxRate
BEGIN
IR-VAL (1.0 - IR) * 100
ETX-VAL ETX
LNK-COST ETX-VAL + IR-VAL
IF LNK-COST > THRESHOLD-VAL THEN
TxRate TxRate - Incr
IF TxRate < LB THEN
TxRate LB
ENDIF
ELSE
TxRate TxRate + Incr
IF TxRate > UB THEN
TxRate UB
ENDIF
ENDIF
nTxRate TxRate
RETURN nTxRate
END
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 25
Wireless Mesh Networks
Chapter 4
IMPLEMENTATION
This chapter contains detailed information about the code level imple-
mentation of CLIA routing metric. Here, we describe the changes and
implementation done, in all the required functions of the existing OLSR
wireless mesh protocol in NS 2.33.
4.1 New Routing Metric CLIA
The CLIA metric of Link l is dened as follows:
CLIA The new routing metric derived here is termed as CLIA It is
the new routing metric which stands for Contention Load and Interference
Aware.It assigns weights to individual links which considers link load, loss
ratio and interference.
Figure 4.1: CLIA
ETX:Expected Transmission Count.
CW: Contention Window.
IR: Interference Ratio.
NL: Number of interferer nodes.
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 26
Wireless Mesh Networks
Cn: Channel Utilization.
CLIAl and CLIAp are routing metric value of individual link and whole
path respectively.
4.1.1 Implementation in OLSR pkt.h and repositories.h
typedef struct OLSR_iface_address {
/// Interface Address
nsaddr_t iface_address_;
/// Link quality extension
double link_quality_;
double nb_link_quality_;
/// Link delay extension
double link_delay_;
double nb_link_delay_;
double nb_sum_link_delay_;//Added by Vinicius
double Nl_CBT_;//Added By Vinicius
inline nsaddr_t& iface_address() { return iface_address_; }
/// Link quality extension
inline double& link_quality() { return link_quality_; }
inline double& nb_link_quality() { return nb_link_quality_; }
/// Link delay extension
inline double& link_delay() { return link_delay_; }
inline double& nb_link_delay() { return nb_link_delay_; }
inline double& nb_sum_link_delay() { return nb_sum_link_delay_; }
inline double& link_CBT() {return Nl_CBT_;}//Added By Vinicius
inline double etx() {
double etx_;
double mult = (double) (link_quality() * nb_link_quality());
double delay = (double) (link_delay() + nb_link_delay());
double BETA;
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 27
Wireless Mesh Networks
if(irVal>=0.9)
BETA = 1; /// MILD-IR Optimization
else if(irVal<0.9 && irVal>=0.5)
BETA = 0.2;
else
BETA = 0.1;
if(chanutil < 0.1) chanutil = 0.1;
printf("IR : %lf\n",irVal);
printf("Nl : %d\n",interfererNodes);
printf("CW : %lf\n",CW);
printf("Cn : %lf\n",chanutil);
switch (parameter_.link_quality()) {
case OLSR_BEHAVIOR_ETT:
return (mult < 0.01) ? 100.0 : ((double) 1.0 / (double)
mult) * (double) delay;
break;
case OLSR_BEHAVIOR_ETX:
if(metricSel == 0 )
{
etx_ = (mult < 0.01) ? 100.0 : (double) 1.0 / (double) mult;
rtMetric = etx_ ;
}
else
{
etx_ = (mult < 0.01) ? 100.0 : (double) 1.0 / (double) mult;
rtMetric = etx_ ;
if(tRate<4096*19) etx_ = (etx_ + interfererNodes) / irVal ;
else etx_ = (etx_ + interfererNodes) / irVal + BETA
* (CW * chanutil) / irVal;
}
return etx_;
break;
case OLSR_BEHAVIOR_MIND:
if(metricSel == 0 )
{
etx_ = (mult < 0.01) ? 100.0 : (double) 1.0 / (double) mult;
}
else
{
etx_ = (mult < 0.01) ? 100.0 : (double) 1.0 / (double) mult;
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 28
Wireless Mesh Networks
if(tRate<4096*19) etx_ = (etx_ + interfererNodes) / irVal ;
else etx_ = (etx_ + interfererNodes) / irVal + BETA * (CW * chanutil) / irVal;
}
return etx_;
break;
case OLSR_BEHAVIOR_ML:
return mult;
break;
case OLSR_BEHAVIOR_NONE:
default:
return 0.0;
break;
}
}
} OLSR_iface_address;
4.1.2 Implementation in mac-timers.h
class MacTimer : public Handler
{
public: .
.
private:.
.
protected:
.
.
double s_idle_period;
double idle_period_;
.
.
}
class BackoffTimer : public MacTimer
{
public: .
.
void pause(void);
void resume(double difs,int state=0);
private: .
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 29
Wireless Mesh Networks
.
};
4.1.3 Implementation in mac-802 11.cc
//Added by Daniel Pereira
#include "ErrorModel80211.h"
/////////////////////////////////////////////////////////////
// Added by Sushmita to support event tracing
#include "agent.h"
#include "basetrace.h"
#include <math.h>
#include <mild.h>
.
.
.
inline void
Mac802_11::checkBackoffTimer()
{
if(is_idle() && mhBackoff_.paused()) {
mhBackoff_.resume(phymib_.getDIFS());
/////////////////////////// MultiRadio-OLSR ///////////////////
//Added by Daniel Pereira and Vinicius Cunha
if (pktTx_ && mhBackoff_.idle_period() >= 0.0) {
struct hdr_cmn *ch = HDR_CMN(pktTx_);
ch->idle_period = mhBackoff_.idle_period();
}
////////////////////////////////////////////////////////
}
if(! is_idle() && mhBackoff_.busy() && ! mhBackoff_.paused())
mhBackoff_.pause();
}
inline void
Mac802_11::transmit(Packet *p, double timeout)
{
/////////////////////////// MultiRadio-OLSR //////////////
float total_period = 0.0;
float idle_period = 0.0;
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 30
Wireless Mesh Networks
///////////////////////////////////////////////////////
tx_active_ = 1;
/////////////////////////// MultiRadio-OLSR /////////
struct hdr_cmn *ch = HDR_CMN(p);
if (mhBackoff_.idle_period() >= 0.0) {
total_period = Scheduler::instance().clock() - macmib_.arrival_time_;
//idle_period = mhBackoff_.idle_period();
//p->txinfo_.cbt = (total_period - idle_period) / total_period;
//ch->cbt = (total_period - idle_period) / total_period;
//ch->idle_period = idle_period;
ch->total_period = total_period;
//printf("[%f] Rx Power idle:%f \n", Scheduler::instance().clock(),
mhBackoff_.idle_period());
if (ch->idle_period > ch->total_period)
printf("[%f] Rx Power idle:%f > total:%f \n", Scheduler::instance().clock(),
ch->idle_period, ch->total_period, ch->ptype_);
}
//////////////////////////////////////////////////
if (EOTtarget_) {
assert (eotPacket_ == NULL);
eotPacket_ = p->copy();
}
.
.
.
.
MAC_MIB::MAC_MIB(Mac802_11 *parent)
{
/*
* Bind the phy mib objects. Note that these will be bound
* to Mac/802_11 variables
*/
parent->bind("RTSThreshold_", &RTSThreshold);
parent->bind("ShortRetryLimit_", &ShortRetryLimit);
parent->bind("LongRetryLimit_", &LongRetryLimit);
parent->bind("ScanType_", &ScanType);
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 31
Wireless Mesh Networks
parent->bind("ProbeDelay_", &ProbeDelay);
parent->bind("MaxChannelTime_", &MaxChannelTime);
parent->bind("MinChannelTime_", &MinChannelTime);
parent->bind("ChannelTime_", &ChannelTime);
/////////////////////////// MultiRadio-OLSR //////////////
arrival_time_ = 0.0;
arrival_lock_ = false;
/////////////////////////////////////////////
}
.
.
.
.
/////////////////////////// MultiRadio-OLSR //////////////
//Added by Daniel Pereira and Vinicius Cunha
bind_bool("cbt_", &cbt_);
bind ("N_", &N_);
//tblCBT = new int [N_];
tblCBT_idx = 0;
tblIR_idx = 0;
tblCBTValue_idx = 0;
//Create bidimentional array to store CBT values
tblCBT = new double*[N_];
tblIR = new double*[N_];
tblCBTValue = new double*[N_];
for (int row=0; row<N_; row++) {
tblCBT[row] = new double[CBT_SMOOTH];
tblIR[row] = new double[IR_SMOOTH];
tblCBTValue[row] = new double[CBT_SMOOTH];
}
.
.
.
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 32
Wireless Mesh Networks
4.1.4 Implementation in OLSR.cc
#include <olsr/OLSR.h>
#include <olsr/OLSR_pkt.h>
#include <olsr/OLSR_printer.h>
#include <olsr/OLSR_dijkstra.h>
#include <olsr/OLSR_parameter.h>
#include <math.h>
#include <limits.h>
#include <address.h>
#include <ip.h>
#include <cmu-trace.h>
#include <map>
#include <mild.h>
/// Length (in bytes) of UDP header.
#define UDP_HDR_LEN 8
//Added by Daniel Pereira
//const int EM80211_MAX_SNR=50;
int pktSize;
double rtMetric;
double maxETT=0.0;
double minETT=999999999999.0;
bool rateEnabled = false;
int metricSel = 0;
//0->ETX and 1->CLIA
4.1.5 Implementation in cbr trac.cc
#include "random.h"
#include "trafgen.h"
#include "ranvar.h"
#include <mild.h>
double tRate;
.
.
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 33
Wireless Mesh Networks
.
double CBR_Traffic::next_interval(int& size)
{
tRate = rate_; /// Mild IR
// Recompute interval in case rate_ or size_ has changes
if(!assign)
{
oldRate=rate_;
assign=1;
incr=4096;
}
.
.
.
if(rateEnabled)
{
double ir,metric;
//if(ir <= 0.1) ir = 0.1;
//else if(ir >= 0.9) ir = 0.9;
//printf("\nIR : %lf\tMETRIC : %lf",ir,rtMetric);
ir = (1.0-irVal) * 100;
metric = rtMetric + ir;
//printf("\nMetric : %lf",metric );
if(metric>140)
{
rate_-=incr;
//if(rate_<oldRate/thresh)
if(rate_< (oldRate * thresh))
rate_=oldRate * thresh;
}
else
{
rate_+=incr;
//if(rate_>oldRate*thresh)
if(rate_>(oldRate*(thresh + 0.7)))
rate_= oldRate * (thresh + 0.7);
}
}
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 34
Wireless Mesh Networks
Chapter 5
Results and Discussion
This chapter contains the results obtained as output of our project and
the analysis of the same. This section evaluates the performance of the
metric CLIA with existing metrics. We have used ns-2 simulator version
2.33 to analyze the proposed routing metric.
5.1 Performance metrics:
The performance of each routing metric is compared using the three
QoS parameters throughput, average end-to-end delay and packet loss
fraction.
Throughput: This is the sum of data packets delivered to the termi-
nals in the network.
Average End-to-End Delay: This is the average overall delay for
a packet to traverse from a source node to a destination node.
Packet loss fraction: The total percentage of transmitted packets
that are lost or dropped in the network during routing from source to
destination.
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 35
Wireless Mesh Networks
5.2 Simulation Environment
We create node topologies using CMU tool setdest to generate large
number of nodes and their movements. The tool uses a random way
point model. We also set up random trac connections of CBR between
nodes using a trac-scenario generator script cbrgen.tcl.We evaluate the
performance of these routing protocols for these topologies.
Figure 5.1: Scenario I
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 36
Wireless Mesh Networks
Figure 5.2: Throughput Vs Packet Rate of Scenario I
Figure 5.3: End to end delay Vs Packet Rate of Scenario I
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 37
Wireless Mesh Networks
Figure 5.4: PDF Vs Packet Rate of Scenario I
Figure 5.5: Scenario II
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 38
Wireless Mesh Networks
Figure 5.6: Throughput Vs Packet Rate of Scenario II
Figure 5.7: End to end delay Vs Packet Rate of Scenario II
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 39
Wireless Mesh Networks
Figure 5.8: PDF Vs Packet Rate of Scenario II
Figure 5.9: Scenario III
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 40
Wireless Mesh Networks
Figure 5.10: Throughput Vs Packet Rate: CLIA vs CLIA-RA
Figure 5.11: Scenario IV
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 41
Wireless Mesh Networks
Figure 5.12: Throughput Vs Packet Rate: CLIA vs CLIA-RA
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 42
Wireless Mesh Networks
Chapter 6
Conclusion and Future Work.
6.1 Conclusion
WMN is a communication network made up of radio nodes orga-
nized in mesh topology. Wireless mesh networks consist of mesh routers
and mesh clients, where mesh routers have minimal mobility and form
the backbone of WMNs. Several routing metrics are proposed for WMN.
But, these routing metrics fail to recognize characteristics such as dier-
ent transmission rates, packet loss rates, and interference.
The proposed metric performs better then ETX and iAWARE. The Rate
Adaptation Mechanism further improves the performance of the proposed
metric.
6.2 Future Work
To implement application based rate adaptation algorithm for spe-
cic applications like FTP, video transmissions.Make a patch for the net-
work simulator so others can learn and continue on.
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 43
Wireless Mesh Networks
Bibliography
[1] DSJDCouto,DAguayo,JBicket,RMorris,Ahighthroughputpathmetricformultihop
wirelessrouting.Wirel.Netw.11,419434(2005).http://dx.doi.org/10.1007/s11276-
005-1766-z
[2] BSadeghi,VKanodia,ASabharwal,EKnightly,Opportunisticmediaaccessformulti
rateadhocnetworks,inProceedingsofthe8thannualinternationalconferenceonMobilecomputing
andnetworking,MobiCom02,ACM,NewYork,2002,pp.2435.http://doi.acm.org/10.1145/
570645.570650
[3] LMa,MDenko,Aroutingmetricforloadbalancinginwirelessmeshnetworks,in21st
InternationalConferenceonAdvancedInformationNetworkingandApplicationsWorkshops,2007,
AINAW07,NiagaraFallsOntario,Canada,21-23May2007,vol.2,pp.409414
[4] MGenetzakis,VSiris,Acontention-awareroutingmetricformulti-ratemulti-
radio
meshnetworks,in5thAnnualIEEECommunicationsSocietyConferenceonSensor,MeshandAdHoc
CommunicationsandNetworks,2008. SECON08,SanFrancisco,California,USA,16-
20June2008,
pp.242250
[5] DMShila,TAnjali,Loadawaretracengineeringformeshnetworks.Comput.Commun
.31(7),1460-1469(2008).http://www.sciencedirect.com/science/article/B6TYP-
4RP0MMN10/
2/8d112fe4f3e
25b28e71fb245e8f23b0e
[6] LTNguyen,RBeuran,YShinoda,Aload-awareroutingmetricforwirelessmeshnetworks,inIEEE
Symposiumon ComputersandCommunications,ISCC,Marrakech,Morocco,6-
9July2008,pp.429435
[7] VBorges,DPereira,MCurado,EMonteiro,Routingmetricforinterferenceandchannel
diversityinmultiradiowirelessmeshnetworks,inAdHoc,MobileandWirelessNet-
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 44
Wireless Mesh Networks
works,vol.5793,ed.
byPRuiz,JGarcia-Luna-Aceves,(Springer,BerlinHeidelberg,2009),pp.5568
[8] LTNguyen,RBeuran,YShinoda,Aninterferenceandloadawareroutingmetric
forwirelessmeshnetworks
Department of Information Science & Engineering, B.V.B.C.E.T., Hubli - 31. 45

You might also like