Published on

IJERA (International journal of Engineering Research and Applications) is International online, ... peer reviewed journal. For more detail or submit your article, please visit

  • Be the first to comment

  • Be the first to like this


  1. 1. Rakesh Kumar, Anjum Sharma / International Journal of Engineering Research and Applications (IJERA) ISSN: 2248-9622 Vol. 2, Issue 4, July-August 2012, pp.906-911 Fault Tolerance for Chain based Hierarchical Data Gathering Protocol for WSN Rakesh Kumar Anjum SharmaAbstract Wireless sensor network nodes are very inaccessible terrains or disaster relief operations. Ontiny in size and their cost is also not very high. the other hand, this also means that sensor networkThey are deployed in any geographical region in a protocols and algorithms must possess selfrandom fashion. During the process of data organizing capabilities. Instead of sending the rawsensing, data gathering and data transmission, the data to the nodes responsible for the fusion, they usecharge of the power unit associated with any node their processing abilities to locally carry out simplegets low, after certain time, i.e., each node has its computations and transmit only the required andlife time. The life time of nodes directly affects the partially processed data. The above describedlife time of the sensor network. As each node is features ensure a wide range of applications forvery low in cost, it is unnecessary and difficult too, sensor networks. Some of the application areas areto recharge them once their energies are health, military, and home. In military, for example,exhausted. Therefore, it is very important to the rapid deployment, self-organization, and faultconserve the power of the nodes so that the life tolerance characteristics of sensor networks maketime of the entire network can be conserved. them a very promising sensing technique for militaryHence the requirement of a power efficient data command, control, communications, computing,gathering protocol is very important to serve the intelligence, surveillance, reconnaissance, andpurpose in wireless sensor network. In the targeting systems. In health, sensor nodes can also beproposed work, it is being tried to change the idea deployed to monitor patients and assist disabledrelating to the data gathering and transmission of patients. Some other commercial applications includethe existing model, as, chain leaders belonging to managing inventory, monitoring product quality, andcertain covering angle will only transmit the monitoring disaster areas.gathered data to the another chain leader of thesame covering angle and we have implemented thebackup chain leader as a fault tolerancemechanism . Our research can provide betterefficiency and resource consumption and furthercan provide good level of fault tolerance.Keywords: Wireless Sensor Node, Cluster, ClusterHead, Fault tolerance1. INTRODUCTION Recent advances in wirelesscommunications and electronics have enabled the Figure 1: Sensor nodes scattered in a sensor fielddevelopment of low-cost, low power, multifunctionalsensor nodes that are small in size and communicate The sensor nodes are usually scattered in a sensorunmetered in short distances. These tiny sensor field as shown in Figure1.Each of these scatterednodes, which consist of sensing, data processing, and sensor nodes has the capabilities to collect data andcommunicating components, leverage the idea of route data back to the sink. Data are routed back tosensor networks. Sensor networks represent a the sink by a multi-hop infrastructure lesssignificant improvement over traditional sensors. A architecture through the sink as shown in Figure1.sensor network is composed of a large number of The sink may communicate with the task managersensor nodes that are densely deployed either inside node via Internet or satellite. The design of the sensorthe phenomenon or very close to it. The position of network as described by Figure1 is influenced bysensor nodes need not be engineered or many factors, including fault tolerance, scalability,predetermined. This allows random deployment in production costs, operating environment, sensor 906 | P a g e
  2. 2. Rakesh Kumar, Anjum Sharma / International Journal of Engineering Research and Applications (IJERA) ISSN: 2248-9622 Vol. 2, Issue 4, July-August 2012, pp.906-911network topology, hardware constraints, transmission protocols utilize position information to relay the datamedia, and power consumption. to the desired regions rather than the whole network.2. ROUTING TECHNIQUES IN WSN 3. RELATED WORK Wireless sensor networks (WSN) consist of In general, three strategies are consideredsmall nodes with sensing, computation, and wireless for the design of data aggregation techniques incommunications capabilities. Many routing, power WSNs. They are cluster based, tree-based, and chain-management, and data dissemination protocols have based . In this section, we only review three chain-been specifically designed for WSNs where energy based routing protocols, PEGASIS, COSEN, andawareness is an essential design issue. Routing Enhanced PEGASIS, and point out their pros andprotocols in WSNs might differ depending on the cons that motivate our research.application and network architecture. Overall, therouting techniques are classified into three categories A. PEGASISbased on the underlying network structure: flat, PEGASIS is a basic chain-based routinghierarchical, and location-based routing. protocol. In which, all nodes in the sensing area areFurthermore, these protocols can be classified into first organized into a chain by using a greedymultipath-based, query-based, negotiation based,QoS-based, and coherent based depending on the algorithm, and then take turns to act as the chainprotocol operation. Many routing, power leader. In data dissemination phase, every node receives the sensing information from its closestmanagement, and data dissemination protocols have upstream neighbour, and then passes its aggregatedbeen specifically designed for WSNs where energy data toward the designated leader, via its downstreamawareness is an essential design issue. Routingprotocols in WSNs might differ depending on the neighbour, and finally the base station. Although theapplication and network architecture. Overall, the PEGASIS constructs a chain connecting all nodes to balance network energy dissipation, there are stillrouting techniques are classified into three categories some flaws with this scheme. 1) For a large sensingbased on the underlying network structure: flat, field and real-time applications, the single long chainhierarchical, and location-based routing. may introduce an unacceptable data delay time. 2)Furthermore, these protocols can be classified intomultipath-based, query-based, and negotiation-based, Since the chain leader is elected by taking turns, forQoS based, and coherent based depending on the some cases, several sensor nodes might reversely transmit their aggregated data to the designatedprotocol operation. Figure 2 shows routing protocols leader, which is far away from the BS than WSNs. This will result in redundant transmission paths, and therefore seriously waste network energy. 3) The single chain leader may become a bottleneck. B. COSEN In contrast to PEGASIS, COSEN is a two- tier hierarchical chain-based routing scheme. In that scheme, sensor nodes are geographically grouped into several low-level chains. For each low-level chain, the sensor node with the maximum residual energy is elected as the chain leader. Moreover, with the low-level leaders, a high-level chain and its corresponding chain leader will be eventually formulated. In data communications, all common (normal) nodes perform a similar procedure as that in PEGASIS to send their fused data, via their respective low-level leaders and the high-level leader, toward the BS. COSEN, compared to PEGASIS,Figure 2: Routing Protocols in WSNs although can alleviate the transmission delay and energy consumption, it still introduces a lot ofIn flat networks all nodes play the same role, while redundant transmission paths, especially for thosehierarchical protocols aim to cluster the nodes so that nodes which are nearest to the BS but would detourcluster heads can do some aggregation and reduction their fused data toward the farther leaders.of data in order to save energy. Location-based 907 | P a g e
  3. 3. Rakesh Kumar, Anjum Sharma / International Journal of Engineering Research and Applications (IJERA) ISSN: 2248-9622 Vol. 2, Issue 4, July-August 2012, pp.906-911C. Enhanced PEGASIS definition of R, different integers (1 … n) are used to indicate distinct angles. In 2007, Jung et al. proposed a variation ofPEGASIS routing scheme, termed as Enhanced  Gθ, R : the group id. Theoretically, byPEGASIS. In their method, the sensing area, centered changing different values of θ and R, theat the BS, is circularized into several concentric sensing area can be divided into n * ncluster levels. For each cluster level, based on the groups. Those are G1, 1, G1, 2, …, G1, n,greedy algorithm of PEGASIS, a node chain is …, Gn, 1, …, Gn,, n.constructed. In data transmission, the common nodesalso conduct a similar way as the PEGASIS to  ni: the node i; the node set N={n1, n2, n3,transfer their sensing data to its chain leader. After …, ni}, where 1≤i≤|N|.that, from the highest (farthest) cluster level to the  cx,y: the id of a chain which was formed inlowest (near to the BS), a multi-hop and leader-by- group Gx,y. the chain set C={c1,1, c1,2,leader data propagation task will be followed. The ….}.EPEGASIS although has considered the location ofthe BS to slightly improve the redundant transmission  lx,y: the leader node id of chain cx,y. Thepath and the network lifetime, there are still some leader set L={l1,1, l1,2, ….}.problems with that scheme. 1) For large sensing  neighbour(ni): the neighbouring nodes ofareas, the node chain in each concentric cluster would ni. The neighbouring nodes mean the nodesstill become lengthy, and thus result in a longer which are locating in the transmission rangetransmission delay. 2) Since the leader node election of a specific node.strategy is same as that in PEGASIS (by takingturns), it did not consider the node’s residual energy.  dis(x, y): the distance between nodes x andAs a node with the least residual energy is elected to y. The BS can be deemed as a specialact as the leader, the network lifetime would be sensor node.significantly affected. 3) While the distribution of B. OPERATION OF CHIRONsensor nodes is not even, the transmission distancebetween two chain-leaders in different cluster levels The operation of CHIRON protocol consistsmight be lengthy, this would consume more energy. of four phases: 1) Group Construction Phase. 2) Chain Formation Phase. 3) Leader Node Election4. CHIRON PROTOCOL Phase. and 4) Data Collection and Transmission Phase. For improving the deficiencies with theaforementioned three schemes, in this section, we 1. Group Construction Phasethus propose an energy efficient hierarchical chain- The main purpose of this phase is ready to divide thebased routing protocol, termed as CHIRON. The sensing field into a number of smaller areas so thatdesign philosophy is described as follows. the CHIRON can create multiple shorter chains toA. Network model and assumptions reduce the data propagation delay and redundant transmission path in later phases. Instead of using Without loss of generality, in our research, concentric clusters as EPEGASIS scheme does, thewe also consider a WSN of n energy-constrained CHIRON adopts the technique of Beam Star tosensor nodes, which are randomly deployed over a organize its groups. After the sensor nodes aresensing field. The BS is located at a corner of the scattered, the BS gradually sweeps the whole sensingsensing area, and equipped with a directional antenna area, by successively changing different transmissionand unlimited power. As a result, the BS can power levels and antenna directions, to send controladaptively adjust its transmission power level and information (including the values of R and θ) to allantenna direction to send control packets to all nodes nodes. After all nodes receiving such control packets,in the WSN. Besides, for easy discussion, we define they can easily determine which group they aresome notations as follows: respectively belonging to. In addition, by the received  R: the transmission range of the BS. For signal strength indication (RSSI), every node can also simplicity, we use distinct integers (1 … n) figure out the value of dis(ni, BS). A grouping to represent various ranges. example with R=1...3 and θ=1..2 is shown in Figure 3.  θ: the beam width (covering angle) of the directional antenna. Also, similar to the 908 | P a g e
  4. 4. Rakesh Kumar, Anjum Sharma / International Journal of Engineering Research and Applications (IJERA) ISSN: 2248-9622 Vol. 2, Issue 4, July-August 2012, pp.906-911 Fig. 3 Grouping example with R=1…3 andθ=1...22. Chain Formation PhaseIn this phase, the nodes within each group Gx,y willbe linked together to form a chain cx,y, respectively.The chain formation process is same as that in Figure 4: The group chains constructed from Figure 1PEGASIS scheme. For each group Gx,y, the node niwith the maximum value of dis(ni, BS) (that isfarthest away from the BS) is initiated to create thegroup chain. By using a greedy algorithm, the nearestnode (to ni) of neighbor(ni) will be chosen to link thenode ni, and become as the newly initiate node innext linking step. The process is repeated until allnodes are put together, and thus finally a group chaincx,y is formed. Figure 2 shows all group chains thatare constructed from the sensing environment ofFigure 23. Leader Node Election PhaseFor data transmission, a leader node in each groupchain must be selected for collecting and forwarding Figure 5: The data transmission flowsthe aggregated data to the BS. Unlike the PEGASISand EPEGASIS schemes, in which the leader in eachchain is elected in a round-robin manner, CHIRON 4. Data Collection and Transmission Phasechooses the chain leader (lx,y) based on themaximum value Res(ni) of group nodes. Initially, in After completed the previous three phases, the dataeach group, the node farthest away from the BS is collection and transmission phase begins. The dataassigned to be the group chain leader. After that, for transmission procedure in CHIRON is similar to thateach data transmission round, the node with the in PEGASIS scheme. Firstly, the normal nodes inmaximum residual energy will be elected. The each group Gx,y transmit their collected data alongresidual power information of each node ni can be the cx,y, by passing through their nearest nodes, topiggybacked with the fused data to the chain leader the chain leader lx,y. And then, starting from thelx,y along the chain cx,y, so that the chain leader can farthest groups, the chain leaders collaborativelydetermine which node will be the new leader for next relay their aggregated sensing information to the BS,transmission round. in a multi-hop, leader-by-leader transmission manner. In order to avoid a longer transmission distance incurred between two chain leaders, and thus result in a great amount of energy dissipation, 909 | P a g e
  5. 5. Rakesh Kumar, Anjum Sharma / International Journal of Engineering Research and Applications (IJERA) ISSN: 2248-9622 Vol. 2, Issue 4, July-August 2012, pp.906-9115. IMPROVED CHIRON SCHEMES changes in angles due to the sensing elements . Due to this sensing the time and power. CHIRON is used to split the sensing fieldinto a number of smaller areas, so that it can createmultiple shorter chains to reduce the data 4 chiron vs improve x 10transmission delay and redundant path, and therefore 2 chironeffectively conserve the node energy and prolong the 1.8 improvenetwork lifetime. In CHIRON routing is done on the 1.6basis of angles. As the routing start from the 1.4clusterhead to destination the path changes in angles 1.2due to the sensing elements . Due to this sensing the y(data)time and power dissipation increases hence the 1CHIRON has not exact output, when we compare to 0.8the real applications. In proposed schemes 0.6communication, routing is done between clusterhead 0.4(CH) to clusterhead (CH) is done directly or as a 0.2straight line. In proposed scheme network is divide 0into two parts, so only two sensor element are 0 500 1000 1500 2000 2500 3000 x(time)present between two clusterheads (CHs). Hence wehave two straight path for routing. Since the number (a)of sensor elements are reduced so sensing time andpower dissipation are reduced hence the 9000 chiron vs improveimprovement in CHIRON is possible. chiron 8000 improve 70006. SIMULATION AND RESULTS 6000 For evaluation the performance and fault y(balance) 5000tolerance of our proposed CHIRON protocol, in this 4000section, we use a simulation tool MATLAB to 3000conduct several experiments 2000A. Simulation environment and parameters 1000 In our simulations, we consider three 0different sizes of sensing area: 100 m*100 m, 200 0 500 1000 1500 2000 2500 3000 x(time)m*200 m, and 300 m*300 m, each is with 100randomly deployed sensor nodes. The BS is located (b)on the corner of sensing field. Every sensor node isinitially equipped with 0.5 joules power. We definethe average delay ass the average required hops, and chiron vs improvethe redundant transmission path as the number of 15 chirondetour hops, for one node transmits its sensing data to improvethe BS, respectively. We also define the simulationround as a duration time in which all sensor nodessent a 2000-bit packet to the BSS. For each 10 y(distributed)simulation scenario, the results are drawn by theaverage value of 10 runs.B. Simulation results 5 Figure 6 shows the simulation results ofaverage propagation delay and redundanttransmission path for two compared schemes. Itcould be seen that the proposed scheme. In CHIRON 0 0 500 1000 1500 2000 2500 3000routing is done on the basis of angles. As the routing x(time)start from the cluster-head to destination the path (c) 910 | P a g e
  6. 6. Rakesh Kumar, Anjum Sharma / International Journal of Engineering Research and Applications (IJERA) ISSN: 2248-9622 Vol. 2, Issue 4, July-August 2012, pp.906-911Figure 6: Improved results for CHIRON.(a) Graph [3] C. Y. Chong and S. P. Kumar, “Sensorfor data on network. (b) Graph for balancing of data Networks: Evolution, Opportunities, andon network. (c) Graph for distribution of data on Challenges,” Proceedings of the IEEE, 91, pp. 1247-1256, 2003. These graph shows the improved result for [4] W. R. Heinzelman, A. Chandrakasan, and H.CHIRON, in which Figure 6.a) shows the data Balakrishnan, “Energy Efficientprocess on the network even when the energy is get Communication Protocol for Wirelesszero. Figure 6.b) shows the balancing of traffic using Microsensor Networks,” Proceedings of thethe imroved CHIRON and CHIRON. Improved 33rd Hawaii International Conference onschemes shows the balancing improvement over System Sciences, 2000.CHIRON. Figure 6.c) shows the distributing of traffic [5] W. R. Heinzelman, A. Chandrakasan, and H.using the imroved CHIRON and CHIRON. Balakrishnan, “An Application SpecificImproved schemes shows the distributing Protocol Architecture for Wirelessimprovement over CHIRON. Microsensor Networks,” IEEE Transaction on Wireless Commun. Vol. 1, Issue 4, pp. 660-670, 2002.7. CONCLUSION [6] E. Altman, T. Basar, T. Jimenez, and N. In this paper, we discuss an efficient Shimkin, “Competitive routing in networkshierarchical chain based routing protocol CHIRON, with polynomial costs,” IEEE Trans.which is suitable for large sensor networks with Automat. Control, vol. 47, no. 1, pp. 92-96,power and time constraints. In our approaches, we 2002.utilize the concept of Beam Star topology to dividethe whole sensing field into a number of smaller [7] J. Choi, C. Conrad, C. Malakowsky, J.areas, so that the CHIRON can create multiple Talent, C.S. Yuan, and R.W. Gracy,shorter chains to reduce the data propagation delay “Flavones from Scutellaria baicalensisand redundant transmission path and provide better Georgi attenuate apoptosis and proteinfault tolerance, thus significantly save network oxidation in neuronal cell lines,”energy. dissipation increases. Hence the CHIRON Biochemica et Biophysica Acta, 1571: 201-has not exact output when we compare to the real 210 (2002).applications. In proposed schemes communication,routing is done between clusterhead (CH) toclusterhead (CH) is done directly or as a straight line.In proposed scheme network is divide into two partsso only two sensor element are present between twoclusterheads (CHs). Hence we have sequentialstraight path for routing. Since the number of sensorelements are reduced so sensing time and powerdissipation are reduced.REFERENCES [1] I. F. Akyildiz, W. Su, Y. Sankarasubramaniam, and E. Cayirci, “A Survey on Sensor Networks,” IEEE Commun. Mag. Vol. 38, pp. 102-114, 2002. [2] J. N. Al-Karaki and A. E. Kamal, “Routing Techniques in Wireless Sensor Networks: A Survey,” Proceedings of the IEEE Wireless Communications, Vol. 11, pp. 6-28, 2004. 911 | P a g e