JOURNAL OF TELECOMMUNICATIONS, VOLUME 13, ISSUE 2, APRIL 2012

18

Node Failure Aware Tree-Clustered Routing Algorithm In WSN

Deepthi Kulkarni, B. P. Patil and M. D. Goudar
AbstractWireless Sensor Networks consist of thousands of tiny nodes having the capability of sensing, computation, and wireless communications. Many routing, power management and data dissemination protocols have been specifically designed, where energy consumption is an essential design issues. Since wireless sensor network protocols are application specific, so the focus has been given to the routing protocols that might differ depending on the application and network architecture. The study of various routing protocols for sensor networks presents a classification for the various approaches pursued. This paper have reviewed and compared protocols like Direct, Pegasis, Leach, TCDGP and an improved node failure aware tree cluster routing algorithm is suggested. This routing scheme has the common objective of trying to get better throughput, end to end delay and to extend the lifetime of the sensor network.The improvements have been obtained in the simulations done using NS2. Index Termswireless sensor networks (WSNs), energy-efficient routing, multi-hop routing protocol, MANETs.

1 INTRODUCTION
Advances in sensor technology, low-power electronics, and low-power radio frequency (RF) design have enabled the development of small, relatively inexpensive and low-power sensors, called micro sensors, which can be connected via a wireless network [1-3]. These sensor nodes (or simply nodes) are usually deployed randomly and densely in hostile environment. They collaborate to observe the surroundings and send the information back to the network manager (or base station) when abnormal events occur. Since battery replacement is not an option for networks with thousands of physically embedded nodes, an efficient energy saving protocol is required to prolong the sensor network lifetime. Generally speaking, more the sensors close to circumstance, the more sensed information is precise when sensor are sensing events. For this reason, sensor nodes always are disposed plenty and densely in the sensing field.
PG Student Dept. of Electronics Engg Maharashtra Academy of Engineering Pune, India. Principal, Marathwada Mitra Mandals, Institute of Technology, Lohgaon. Pune, India.

Sensor nodes through the collaborative effort send many kinds of the environment information to the remote sink. After sink aggregating and computing data, sink will convey data to external network by way of Internet or satellite network. It is not easy to supply large power to sensor node because the battery is restricted on the nodes volume and it does not have the problem in MANETs. For above-mentioned reasons, a lot of routing protocols are purposed to improve the power consumption in wireless sensor networks [4]. Network lifetime can be defined as the time elapsed from the network operation starts until the first node (or the last node) in the network depletes its energy (dies). Energy consumption in a node can be due to either useful or wasteful operations. The useful energy consumption includes transmitting or receiving data messages, and processing query requests. On the other hand, the wasteful consumption can be due to overhearing, retransmitting because of harsh environment, and dealing with the redundant broadcast overhead messages, as well as idle listening to the media. In this paper we shall compare various protocols used for data gathering in WSN and concentrate

HOD, Electronics, Maharashtra Academy of Engg. Alandi, Pune India. on parameters such as node death, end to end delay and shall propose a new node failure aware algorithm to overcome all the disadvantages of above algorithms.

2 Overview of WSN
2.1. Features and Requirement

2012 JOT www.journaloftelecommunications.co.uk

JOURNAL OF TELECOMMUNICATIONS, VOLUME 13, ISSUE 2, APRIL 2012

19

Application specific wireless sensor networks consist of hundreds to thousands of low-power multi-functioning sensor nodes, operating in an unattended environment, with limited computational and sensing capabilities, they demand following requirements A sensor node should be inexpensive. Data gathering protocol should be efficient enough to give longer life to the network. Nodes should be able to form a network automatically without any external configuration. Sensor nodes should be able to work together and aggregate their data in a meaningful way. There may be many other application specific components beside these. Like in LEACH [5] and PEGASIS [6], we also use first order radio model for message transmission and reception. Fig.1: Architecture of WSN

2.2. System Architecture A sensor network can, in practice, be composed of tens to thousands of sensor nodes which are distributed in a wide
area. These nodes form a network by communicating with each other either directly or through other nodes. One or more nodes among them will communicate with the user through the Base Station, either directly or through the existing wired networks. Fig.1 shows a typical architecture of a sensor network in which sensor nodes are shown as small circles. Each sensor node typically consists of the five components such as sensor unit, analog digital converter (ADC), central processing unit (CPU), power unit, and communication unit. ADC is a translator that tells the CPU what the sensor unit has sensed, and also informs the sensor unit what to do. Communication units task is to receive command or query from, and transmit data from CPU to outside world. CPU is the most complex unit. It interprets the command or query to ADC, monitors and controls power if necessary, processes received data, computes the next hop to the sink, etc.

3. Literature Survey:
Wireless Sensor Networks (WSNs) can gather the sensed information by hundreds or even thousands of sensing nodes and transmit them to the sink. At present, there are many routing methods in the wireless sensor network.The primary three types will be introduced as following:

3.1 Directed Diffusion Protocol It uses the easiest way that sensor nodes transmit the sensed data to sink directly [7]. Using this way is very
simple, but it will have a serious problem. When a farther sensor node transmits the data, it will spend more energy than the closer one. Therefore, it is desirable to make these nodes as energy-efficient as possible and to rely on their large numbers in order to obtain high quality results. Likewise, the sensor network routing protocols must be designed to achieve fault tolerance in the presence of individual node failures while also minimizing energy consumption. Moreover, since the limited wireless channel bandwidth must be shared by all the sensors in the network, routing protocols for these networks should be able to perform local collaborations in order to reduce the bandwidth requirements. Eventually, the data being sensed by the nodes in the network must be transmitted to a control center (i.e., the sink) or base station where the end sensor nodes can access the data.

2012 JOT www.journaloftelecommunications.co.uk

JOURNAL OF TELECOMMUNICATIONS, VOLUME 13, ISSUE 2, APRIL 2012

20

3.2 LEACH Low-Energy Adaptive clustering Hierarchy (LEACH) protocol is representative cluster-based of routing protocols. In a sensor field, sensor node senses data and sends data to the sink that called round. The working procedure for LEACH will be finished in a single round. Before gathering the sensed data for each round, the large number of sensor nodes is divided into several clusters. A cluster head is randomly chosen through self-organization. Each cluster head is in charge of gathering the sensed data from the sensor nodes in the cluster. The cluster head will aggregate the received data and send it directly to the sink. After the sink has received all of the data from the cluster heads a round has ended.

standard tree traversal algorithm. They go into the data transmission phase after building the tree. All the leaf nodes will start sending the sensed data towards their parent nodes. The parent nodes will collect the received data with their own data. Then send the collected data to their parent. The transmission process will be repeated until all the received by the root node. After root node aggregating data, it send collecting data to sink directly. The process will go around until the root node dead. WSN will re-select a new root node. Root id number would be j+1. Then do the initial phase again like above. The tree path will not change until the root node dead. TREEPSI and PEGASIS are using the same way to transmit data from leaf node to chain/root head. The length of path form end leaf node to root/chain node in TREEPSI is shorter than PEGASIS. The data will not send data for a long path. For this reason, TREEPSI can reduce power consumption less in data transmission than PEGASIS. The TREEPSSI has better performance about 30% than PEGASIS. It still has a problem that restriction on the binary tree algorithm, the path has made a detour in the topology.

3.3 PEGASIS Power-Efficient Gathering in Sensor Information Systems (PEGASIS) is based on chain-based protocol and differ from
LEACH. This proposal is building all sensor nodes to form a chain according to Greedy algorithm that the sum of edges must be minimum in wireless sensor networks. At the initial phase before each round, they must choose a chain head. The N represents the number of nodes and all the nodes use the natural number from 1 to N. Then WSNs utilize the i = j mod N to choose chain head. If it is equal to zero, then choose N. The two end-point of the chain will start send sensed data to the parents nodes for forwarding data to the chain head. All the nodes in the chain only transmit data to its neighbor. Each edge only sends or receives data one time, after the chain head received the two children nodes, it will aggregate the data and transmit the collecting data to sink directly.

3.5 Tree- Cluster Data Gathering Protocol (TCDGP) To reduce power dissipation, a novel TCDGP is proposed
[9] to combine cluster-based and tree-based protocols. A sink can obtain location and energy information about all sensor nodes using two methods. One method is recorded in the sink at the initial state as the nodes are deployed. The other method is that the sink can broadcast to the whole network and then receive return messages from the sensor nodes. During WSN initiation there are no routing mechanisms after the sensor nodes are deployed. Thus, for the proposed protocol, the mechanism is established during the initial state. There are two steps in the deployment phases, which are cluster establishment and tree set-up. After the data aggregation, the proposed protocol will judge the threshold value of the remnant energy. During cluster establishment, the main idea is that the sink exploits each sensor nodes localization and energy to compute the adaptive cluster heads for the WSNs. After selecting cluster heads, the heads will build their own clusters. Therefore, the first step is to choose the cluster heads, and the second step is to form clusters, then accordingly cluster based tree is formed.

3.4 Treepsi Tree-based Efficient Protocol for Sensor Information (TREEPSI) [8] is tree-based protocol that is different from
abovementioned protocols. Before data transmission phase, WSNs will select a root node in all the sensor nodes. Set the root identify id=j. There are two ways to build the tree path. One is computing the path centrally by sink and broadcasting the path information to network. The other can be the same tree structure locally by using a common algorithm in each node. At the initial phase, root will create data gathering process to the children nodes using any

2012 JOT www.journaloftelecommunications.co.uk

JOURNAL OF TELECOMMUNICATIONS, VOLUME 13, ISSUE 2, APRIL 2012

21

4 PROPOSED EFFICIENT PROTOCOL

The nodes are organized into tree-cluster network and data gathering takes place. The data communication and network operation causes energy depletion in the sensor nodes. The schemes for failure detection and cluster recovery are activated in the event of failures due to energy-drained nodes. In this paper, the maintenance and recovery of the cluster structure in the event of node failures is termed as failure recovery. Further detail elaboration of failure detection and failure recovery mechanisms can be seen. Fig.2

4.1 Failure detection In this section, we discuss the method to detect energy failures in the nodes and report the same to the respective
members of the clusters. This detection is essential for the cluster members as they have to invoke the mechanism for the repair and recovery of those failures so as to keep the cluster connected. Every node has a record of its balance energy. The nodes in each cluster send their energy status as a part of the hello_msg, to their first hop members including their parent. The hello_msg consists of the location (x and y coordinates), energy and node ID. This hello_msg conveys the current energy status of the node. When the node is failing, it sends the failure report message fail_report_msg to its parent and children. A node is termed as failing when its energy level drops below the threshold value, Eth. The threshold value, Eth, is the energy required to transmit D number of l-bit messages across a distance equal to the transmission range. In Fig. 2, let us assume that node 7 is failing, and then it sends a fail_report_msg to node 3, its parent and node 10 its child. Here we deal with failures related to energy exhaustion, and therefore we assume that the failing node can send the failure report to its immediate hop members before it dies completely. This information of the failure report is an indication to start the failure recovery process. The parent and children of the failing node are sufficient to invoke the failure-recovery mechanism. Therefore energy is saved by not allowing all the nodes in the cluster to detect a failure. This is the method by which all the nodes in the cluster know about the failure of its first hop members and the corrective action is taken by only those nodes that have the information.

4.2) Failure recovery In this section, we discuss the mechanisms for failure recovery. The failure recovery here refers to the connectivity recovery after the node has failed. The node failures discussed here is confined to failure due to energy exhaustion. The failure-recovery mechanisms are performed locally by each cluster. When a node fails, the failing nodes parent and children take appropriate action to connect the cluster and bridge the gap formed by the failing node. This paper proposes four types of failure mechanisms depending on the type of node in the cluster. The nodes in the cluster are classified into four types, boundary node, pre- boundary node, internal node and the head node. Every node has a different mechanism for failure recovery. Common routines followed by recovery algorithms:

4.2.1. Failure reporting A node is considered failing if its energy falls below the threshold energy. A failure report message, fail_report_msg,
is sent by the failing node to its parent and children. This helps the children to realize that they need to search for another suitable parent for further operation. Once the parent receives the failure report, it ignores the failing node for further data transactions and considers it a non-active member.

4.2.2. Procedure for finding a suitable healthy parent: A

join_request_mesg is sent by the healthy child of the failing node to its neighbours. All the neighbours within the transmission range respond with a join_reply_mesg/ join_reject_mesg message. The healthy child of the failing node then selects a suitable parent by checking whether the neighbour is not one among the children of the failing node and whether the neighbour is also not a failing node. If the healthy child is a boundary node, it first searches for a

2012 JOT www.journaloftelecommunications.co.uk

JOURNAL OF TELECOMMUNICATIONS, VOLUME 13, ISSUE 2, APRIL 2012

22

parent within a cluster, if not successful, it then searches for a parent outside the cluster. While searching for a parent, it checks whether the supportable degree of the neighbour is within the limit D, if the parent is of the same cluster. If the parent is from different cluster, the supportable degree of the neighbour must be within the limit D and the cluster size limit also must be within S. If a healthy child is an internal node, it searches for a suitable parent inside the cluster only. If a suitable parent is found, then the healthy child node then attaches itself to the chosen parent. The cluster_info_mesg is exchanged if the chosen parent is from a different cluster. The cluster parameters of the child are updated to that of the new chosen parent through update_mesg and data transmissions then follow the new paths. The failing node is then left with the original parent and its children are all allocated different parents to keep their data transmissions uninterrupted.

Fig. 3: Direct Routing

The various failure recovery algorithms for a boundary node, pre-boundary node, internal node and a head node is been detailed in [10].

5 RESULT ANALYSIS
In our work, the new node failure aware data gathering protocol for wireless sensor network has been introduced. Detailed simulations are done on NS2 considered with 30 nodes placed at random and chosen parameters are End-to- end delay, Node death failure and throughput. X-axis gives the time of packets sent and y-axis gives packets lost. Result 1: Compared Direct, Pegasis, Leach and TCDGP node death failure. Fig. 4: Pegasis Routing

Fig. 5: Leach Routing

2012 JOT www.journaloftelecommunications.co.uk

JOURNAL OF TELECOMMUNICATIONS, VOLUME 13, ISSUE 2, APRIL 2012

23

Fig. 6: TCDGP Routing These results show us the Node Death in various protocols. Finally the graphs indeed explain that TCDGP have less node death as compared to other protocols, more data is sent with less node death failure. Hence TCDGP excels other protocols, the further comparison would be between TCDGP and Recovered TCDGP which is proposed in this paper considering other parameters. Result 2: Various parameters compared between Normal TCDGP and Recovered TCDGP protocol.

Fig. 8: Delay Analysis

The results showed above shows how the new node failure aware routing algorithm excels TCDGP in case of node death and delay. Fig.7 shows node death failure analysis of new algorithm, which explains that at the beginning due to more data sent, maximum node death is seen at the beginning of both the algorithms but later as time increases, node death of new algorithm is nil and maximum data is sent. Fig.8 shows delay of new algorithm has lessen 6 times than TCDGP. This delay is end to end delay i.e the time taken to send data from node to sink or we can say source to destination.

Fig. 7: Node Death Failure Analysis

Fig. 9: Throughput Analysis

2012 JOT www.journaloftelecommunications.co.uk

JOURNAL OF TELECOMMUNICATIONS, VOLUME 13, ISSUE 2, APRIL 2012

24

Fig.9 gives the throughput analysis of the nodes in proposed algorithm which is less than TCDGP protocol, i.e. this algorithm gives a good node failure and less delay analysis of nodes but couldnt send more data than TCDGP algorithm. VI.Conclusion: From the detailed comparison of various algorithms this paper found that TCDGP excels others, but this paper proposed a new way to detect and recover failure of nodes taking place during data aggregation. A new node failure aware tree clustered routing algorithm though excels in node death failure and also decreases end to end delay but have sent less number of packets in each round i.e. throughput is less than normal TCDGP protocol. Hence this method can be applied where throughput is not a major issue but see that the delay and death of nodes is less. REFERENCES

[10] G. Venkataraman S. Emmanuel S. Thambipillai Energy-efficient cluster-based scheme for failure management in sensor networks IET Commun., 2008, Vol. 2, No. 4, pp. 528537 Deepthi Kulkarni is currently pursuing Masters of Engineering degree in Electronics Engineering from Maharashtra Academy of Engineering Pune, India. Her Current research interests include Wireless Sensor Network and NS2.

Amritsar India in year 2000. Presently working as Principal, Marathwada Mitra Mandals Institute of Technology, Lohgaon, Pune, India. He has published 85 plus research papers in various international and national referred journals and conferences. He is recipient of Sir Thomas Ward memorial Medal from the Institution of Engg., Calcutta, for his best paper in E & T Journal div. for the year 1999- 2000. His area of Interest includes Communication and Wireless Sensor Network. Dr. B. P. Patil received his Bachelor of Engineering degree in Electronics Engg. From Amravati University, India in 1990 and Masters in engineering degree from NIT Allahabad, India in 1996. He received Ph.D. in Electronics Technology from Guru Nanak Dev Univ.,

[1] R. Min, M, Bhardwaj, S. Cho, A. Sinha, E. Shih, A. Wang, and A. P. Chandrakasan, Low Power Wireless Sensor Networks, Proceedings of International Conference, Bangalore, India, January 2001. M D Goudar obtained his Graduate and Post [2] J. M.Rabaey, M. J.Ammer, J. L. da Silva Jr., D. Patel, S. Roundy, Graduate Engineering degree in the year 1987 and Pico Radio supports ad hoc ultra low power wireless networking, 2004 in the discipline of Electronics & Communication IEEE Computer, Vol. 33, pp. 42-48, July 2000 and Electronic Design Technology. He is presently [3] K. Sohrabi, J. Gao, V. Ailawadhi, and G. Pottie., Protocols for working as Associate Professor and Head of the selforganization of a wireless sensor network, IEEE Personal Department at Maharashtra Academy of Engineering, Alandi (D), Communications, Vol. 7, No. 5, pp. 16-27, October 2000. [4] Q. Jiang and D. Manivannan, Routing Protocols for Sensor Pune, India. He is perusing Ph.D. from Indian School of Networks, IEEE Consumer Communications and Networking Mines. Conference, 2004. [5] W. Heinzelman, A. Chandrakasan, and H. Balakrishnan, Energy- Efficient Communication Protocol for Wireless Microsensor Networks, Proc. Hawaii Conf. System Sciences, Jan. 2000. [6] S. Lindsey, C. Raghavendra,Pegasis: Power-Efficient gathering in sensor information systems, In: Proc. of the IEEE Aerospace Conference, 2002. Pp.1-6. [7] Chalermek Intanagonwiwat, Ramesh Govindan, Deborah Estrin, John Heidemann, Directed Diffusion for Wireless Sensor Networking IEEE/ACM transactions on networking,vol 11,no.1,Feb 2003 [8] S.S. Satapathy and N. Sarma , TREEPSI: tree based energy efficient protocol for sensor information ,Wireless and Optical Communications Networks 2006, IFIP International Conference 11-13 April 2006 . [9] Yun-Sheng Yen, Kai-Chun Huang, Han-Chieh Chao*, and Jong Hyuk Park TREE-CLUSTERED DATA GATHERING PROTOCOL (TCDGP) FOR WIRELESS SENSOR NETWORKS Journal of the Chinese Institute of Engineers, Vol. 32, No. 7, pp. 1025-1036 (2009) 2012 JOT www.journaloftelecommunications.co.uk