IMPLEMENTATION OF SECURITY PROTOCOL FOR WIRELESS SENSOR

IMPLEMENTATION OF SECURITY PROTOCOL FOR WIRELESS SENSOR

• 1.
  International Journal onComputational Science & Applications (IJCSA) Vol.5, No.4, August 2015 DOI:10.5121/ijcsa.2015.5408 87 IMPLEMENTATION OF SECURITY PROTOCOL FOR WIRELESS SENSOR NETWORK USING ZONE BASED CLUSTERING ALGORITHM Amit Singh1 ,Khushbu Babbar2 and Apeksha Malik3 Department of Computer Engineering,Banasthali University, Jaipur, Rajasthan ABSTRACT Intrusion Detection is one of the methods of defending against these attacks. In the proposed a security protocol for homogeneous wireless sensor network; network with all nodes are of same type. Clustering is used to improve the energy efficiency. Zone-Based Cluster Protocol (ZBCA) is used for selection of cluster head which is effective in scalability and energy consumption. Single hop technique is used for communication within normal nodes and cluster head to base station. Simulation of proposed algorithm is performed in MATLAB. Sleep Deprivation Attack has been analyzed where attacker changes the environmental values by an artificial event. Attacker produces an event in environment due to which nodes have to sense the environment more than once in the same round that increase the power consumption of the node. This interrupt reduces the network life time as nodes are not allowed to go in sleep mode and they are not able to perform their function of data collection and reporting to Cluster head and Base Station properly. Proposed protocol identifies this attack and prevents it from happening by isolating the attacker node. KEYWORDS Wireless Sensor Network, Security, Intrusion Detection System, Sleep Deprivation Attack, ZBCA. 1.INTRODUCTION The popularity of small handheld devices likes laptops, cell phones, PDAs, GPS devices, and intelligent computing devices are increasing day-by-day. This made the things cheaper, more mobile, more distributed, and more pervasive in daily life of human being. In this scenario, the emergence of Wireless Sensor Networks (WSNs) is essentially toward the miniaturization and ubiquity of computing devices. Wireless sensor networks are found suitable for applications such a surveillance, agriculture, traffic management, disaster detection etc. Sensor networks are composed of thousands of resource constrained sensor nodes and also some resourceful base stations are there intelligent decision making Sensor nodes are constrained in terms of limited battery power, cost, memory limitation and limited computational capability. All nodes in a network communicate with each other via wireless communication. Security is considered as measure of Quality of Service (QoS) of sensing the accuracy of information gathered about events occurring in environment by sensors in sensor network. WSN are deployed in harsh and hostile terrain so are at high risk of physical distortion. Because of this intrinsic nature of networks, WSN becomes vulnerable to many security attacks. Most of the attacks against security in wireless sensor networks are caused by the insertion of false information by the compromised nodes within the network. A system is required for defending the inclusion of false reports by compromised nodes. Many security mechanisms like authentication, key exchange and security
• 2.
  International Journal onComputational Science & Applications (IJCSA) Vol.5, No.4, August 2015 88 routing have been proposed but they cannot deal with providing security towards many attacks. An IDS has provided with the best solution for addressing wide range of security attacks in WSN. IDS help in identifying and reporting unauthorized network activities and deny them access from network resources. Rule Based IDS has been used in our protocol which identifies any possibility of attack by comparing the occurring network patterns with existing attacking patterns or rules. There are many types of attacks but we are working on Sleep Deprivation Attack because it’s a great threat to lifetime of sensor networks as it prevents the node from going into sleep mode due to which batteries drain early and sensor nodes to die early. Therefore it’s necessary to understand this attack very carefully. Proposed protocol analyses the effects of the Sleep Deprivation Attack and provides solution by identifying and isolating the node. In our earlier analysis, Sleep Deprivation attack has been analyzed in both heterogeneous and homogeneous networks. Zone Based Clustering Protocol (ZBCA) has been introduced as clustering algorithm to increase the network lifetime in terms of energy and scalability. Simulations have shown that ZBCA balances the energy consumption by each sensor node deployed in the network 2. PROBLEM STATEMENT Wireless Sensor Networks are more vulnerable than wired applications because WSN are easily accessible as they are not closed systems. So data transferred by WSN can easily be hacked as compared to that by wired networks. Also WSN are limited in terms of computation, power and memory resources which make them susceptible to various attacks. Thus it is required that proper security mechanisms like encryption and authentication should be introduced in systems to provide proper security. But they cannot be easily implemented on WSN due to their intrinsic nature thus allowing an attacker to distribute malicious code and reprogram the deployed sensor nodes with their own code. We have proposed a security protocol for WSN against Sleep Deprivation Attack in network. In this proposed security protocol, LEACH algorithm has been used as the Energy model. Low Energy Adaptive Cluster Hierarchy – based protocol (LEACH) is a cluster based algorithm. This protocol is a round based protocol where in each round a different set of nodes called Cluster Heads (CH) are selected on some probability basis. The cluster heads aggregate and compress the data and forward it to the base station, thus it extends the lifetime of major nodes. Also most of the nodes that are not in use go into Sleep State. LEACH assumes that all nodes are in wireless transmission range of the base station, which is not the case in many sensor deployments. We have done simulation of Sleep deprivation attack in both homogeneous and heterogeneous environment. This simulations show the performance of network when attacker gains access to nodes in the network and introduces malicious node. Generally these intrusions go unnoticed and cause failure of the normal working of the network. Hierarchical architecture has been implemented in our protocol where sensors organized themselves into clusters and each cluster has a cluster head which process, aggregate and forward the information collected from its cluster members, to a potential layer of clusters among themselves toward the base station. Also Rule- Based IDS is used in the network to identify for intrusions. We have used the rule that only CH can send control packet to its own CM. so if any node receives control packets from node other than its CH and that too more than one time, node identifies the sending node as attacker node. This rule in our proposed protocol detects malicious node and isolates it from the network to minimize the effect on normal working of network. 3.PROBLEM FORMULATION Sleep Deprivation Attack or Denial of sleep attack is an attack where an intruder forces the sensors to remain awake so that they waste their energy thus resulting in large power consumption by limited power sensor nodes. So the nodes die early. This attack is a great threat to lifetime of sensor networks as network lifetime is considered as the critical aspect of applications in WSN. Also the nodes located far from the BS die quickly since they dissipate more energy in
• 3.
  International Journal onComputational Science & Applications (IJCSA) Vol.5, No.4, August 2015 89 transmitting data packets. Simulation has been performed in homogeneous sensor network in which sensors are deployed in uniformly distributed manner. In our security protocol, zone based clustering has been used to increase the network life time. The attacking node is identified and isolated. Clustering is used as a key tool for the energy efficient protocol, and in this protocol network is organized in clusters. Such a sensor network contains two types of nodes; cluster head and cluster members. All the data processing such as data fusion and aggregation are local to the cluster. Cluster heads change over time in order to balance the energy dissipation of nodes. Cluster member sense the data and send to cluster head and cluster head sends the data to BS. WSN are usable as long as they can communicate sensed data to processed node. Since sensing and communication are important activities and they consume energy so power management and along with security can effectively increase the network lifetime. 4.IMPROVING WSN USING ZONE BASED CLUSTERING ALGORITHM (ZBCA) Many protocols have been proposed for other wireless networks like mobile or ad- hoc. But these protocols cannot be used directly on wireless sensors due to their resource constraints like limited battery power, computational speed, and human interface of node device and density of nodes in network. So, there is a need of a new protocol scheme which enables efficient use of energy at individual sensor nodes to enhance the network survivability. Clustering has been used as data analysis technique for many applications that require scalability to hundreds or thousands of nodes. It also prolongs the network lifetime. The energy consumption rate for sensors in a WSN varies greatly depending upon the protocols that have been used for consumption in a network. Also the efficiency of WSN depends upon the distance between node to base station and performance of clustering algorithm depends upon the selection of cluster heads that in turn form clusters and control cluster members efficiently. ZBCA is a clustering method for WSN to balance energy consumptions for all sensor nodes in a network field. In this method a network field is divided into several zones each of equal size and each zone is allocated with a cluster head. This type of allocation distributes the cluster heads evenly over entire network field. Each of the sensor nodes transmit data to the nearest cluster head that may belong to its own zone or neighboring zones only. Each cluster head aggregates data and send it towards sink. Thus ZBCA is an energy efficient homogeneous clustering head selection algorithm for WSN that reduces the number of communication between the sensors nodes for cluster head selection so that the energy consumption for cluster head selection can be reduce. 5.NETWORK AND ENERGY PARAMETERS 5.1.Energy Consumption: Power consumption in proposed selection method for cluster head is E consume= Energy to send JOIN request to all neighboring zone nodes in network +Energy for the reception of data for all nodes in cluster +Send the CH information to Base station. = (ETX* ctrPacketLength * Distance from selected CH and all other node in the network) +(ERX*ctrPacketLength *N) + EDA+(ETX * ctrPacketLength + Emp* ctrPacketLength) 5.2 Performance Criteria: There are many performance criteria to evaluate working of WSN like Coverage, Network Lifetime, Miss Ratio, Energy cost, Transmission Delay Time etc. Network Lifetime has been used as the performance criteria in our round based algorithm. Network Lifetime is the number of rounds in which the first node dies. In the random selection method, cluster head selection in each round is done on random basis. When a node becomes dead in the network, it will not be the part of the network. We aim at decreasing the round number when first dead node is found thus increasing the network lifetime.
• 4.
  International Journal onComputational Science & Applications (IJCSA) Vol.5, No.4, August 2015 90 6.PROPOSED APPROACH FOR CLUSTERING USING GENETIC ALGORITHM Proposed security protocol works on homogeneous environment of wireless sensor network and clustering architecture ZBCA is introduced to improve the network lifetime. This algorithm works in rounds. Each round consists of Setup and Transmission phase. Setup phase selects the cluster head on random basis and nodes associate with the cluster head which has minimum distance from that node. Transmission phase includes the process of sensing data and sending to its cluster head. Cluster Heads maintain an array of previous value for all of its cluster members. They perform aggregation and compression functions on sensed data depending upon these previous values and accordingly send it to Base Station. If they find any node sending many control packets in a round, that node is assumed as malicious node and Base Station disowns that malicious or advisory node from the network. 6.1.Network Preliminaries Number of nodes in network is 100.·Sensor nodes are uniformly distributed throughout the region. The network is homogenous and all nodes have same initial energy. All sensor nodes use non-renewable batteries so in order to preserve energy sensor nodes goes in and out of low- power sleep state. The base station (BS) is located inside the network.BS and all sensor nodes ate not mobile. The BS has stored information about the location of each node. All nodes can send data to BS. Cluster heads are randomly chosen in each zone provided that those nodes are not malicious or dead nodes. Cluster Head performs data compression and aggregation. Transmission and reception energy is same for all nodes. Energy of transmission depends upon signal strength that is in terms of the distance (source to destination) and data size. No cluster head will ever become malicious node. Sensed data within range of 0-5 is considered as insignificant change in physical environment so is not send to CH. 6.2.Phases in Proposed Algorithm Proposed algorithm is f or homogeneous environment of wireless sensor networks. The proposed algorithm is work in round. 6.2.1.SETUP PHASE In Setup phase zones are created on basis of geographical area and cluster head is selected for each of the zone. Then re-clustering takes place on basis of minimum distance. The objective o f the process to find the cluster that can minimize the total network energy consumption and provide proper coverage to get maxim um security. Step1: Creating Zone • BS create zone on the basis of geographical area and zoneID is assigned to each zones. In our protocol we divide the network in 9 rectangular. • A random node from the zone is selected as cluster head
• 5.
  International Journal onComputational Science & Applications (IJCSA) Vol.5, No.4, August 2015 91 Figure 1. Network area divided into 9 rectangular zones called grid and Cluster Head (CH) is selected in each grid randomly Step2: Cluster Formation • Selected cluster head s end the JOIN request to the nodes lie in the zone itself and to the neighbor’s zone. • On the basis of the distance the nodes select one cluster head whichever is most near to it. The nodes send acknowledgement to cluster head. And final cluster formation is done. Figure 2. CH sending JOIN request to its own zone and neighboring zone nodes and re-clustering is done 6.2.2.TRANSMISSIO N PHASE Nodes send the sensed data to the Cluster Head and CH performs aggregation and compression on data and send it to Base Station. Step 1: Each node generates random value. Let these random values represent the data sensed like change in temperature or humidity etc. If the sensed data lies in range 5-10 then only data are s end to the cluster head Step 2: Each cluster head consists of an array for its cluster members w hich consists of previous values. Incom ing sensed data is compared with their respective values and accordingly send to base station. Miscellaneous Node Detection and Isolation During transmission phase, any round can have malicious node in them. If a round does not have any such a tacking node then nodes perform their operation of sensing data and forwarding it to base station via cluster head. But if the round is found with a malicious node in it, nodes instead
• 6.
  International Journal onComputational Science & Applications (IJCSA) Vol.5, No.4, August 2015 92 of sending sensed data, forward information about this malicious node to base station via cluster head. • Select any round on random basis to introduce malicious node. • Select any random node from the deployed nodes as miscellaneous or attacking node. We have assumed that cluster heads will never be selected as malicious node because if cluster head themselves become attacking node then they will not inform Base station about their real identity. The attacking node sends more than one control packet to all the node s that lie in its range within same round. This is Sleep Deprivation attack where the attacking node is not allowing nodes to go into sleep mode and forcing them to sense false information. • Receiving nodes compare their own cluster head id with sending node id to ensure about the identity of sending node. • If sending node is found not be Cluster Heads of receiving node, sending node is assumed as malicious node and this information is send to their Cluster Head. • Cluster Head forwards this information to Base Station. • Base Station disowns such malicious node from the network and inform all Cluster Heads about this node • This disowned node will never become the part of network either in form of normal or cluster head nodes. So, our proposed protocol identifies malicious node and isolates it from the network. Now this attacking node can no longer cause any effect in the network. 6.3.Pseudo-code of the Proposed Approach: BEGIN 1: specify number of nodes (n); 2: Einit(s)= Eo, s=1,2,….n; (I) SETUP PHASE 1: Create 9 rectangular grids; 2: Select Cluster Heads using ZBCA; 3: Send JOIN request to all nodes of neighboring zones; 4: Receive ACK from all nodes of neighboring zones; 5: Cluster formation done. Send information to Base Station. (II) TRANSMISSION PHASE 1: if Cluster Member receives a ctrl packet 2: if CH_ID== ID then //ID= sender node’s ID 3: No malicious node detected; 4: else SendToCH (ID); // nodes inform CH about malicious node 5: SendtoBS (ID); // CH inform BS about malicious node 6: ISOLATE (ID) and inform all CH;// malicious node disowned from the network 7: end if; 8: else
• 7.
  International Journal onComputational Science & Applications (IJCSA) Vol.5, No.4, August 2015 93 9: n=sending data; 10: if (0<n<5)// data 0-5 is negligible amount of physical condition sensed by sensors 11: Send no data; 12: else if (5<n<10) // data 0-5 is negligible amount of physical condition sensed by sensors 13: SendtoCH (n); 14: end if; 15: if (CH(s) = TRUE) 16: if (previousData = 0 | previousData = 1 & n = 1)// Data Aggregation 17: Receive (n); 18: previousData=1; 19: else if (previousData = 0 | previousData = 1 & n = 0) 20: Receive no data; 21: previousData=0; 22: end if; 23: TranstoBS(Data);// aggregated data is send to BS 24: end if; 25: end if; 26: } //round is complete 7.SIMULATION PARAMETERS Simulation Parameters : following simulation parameters are used to perform simulation. Table1: Simulation parameters 8. SIMULATION RESULTS When simulation has been done, four scenarios have been observed. These scenarios show the simulation run as well as comparisons made between various architectures used in the proposed algorithm. They are as follows: Scenario 1: This scenarios show that the malicious nodes that are introduced in the system are detected by the nodes lying in range of attack and are isolated by the BS. ZBCA has been implemented in the network which divides entire network in equal size rectangular zones called grids and randomly selects CH in which grid. Following figure shows the simulation run. Parameter Values Simulation Round 1000 Number of rounds 100+1(Node+BS) Topology Size 200 X 200 Nodes Distribution Nodes are Randomly Distributed Energy for Transmission (ETX) 50*0.000000001 Energy for Reception (ERX) 50*0.000000001 Energy for Data Aggregation (EDA) 5*0.000000001
• 8.
  International Journal onComputational Science & Applications (IJCSA) Vol.5, No.4, August 2015 94 Figure 3. Simulation Run showing Grid architecture with 20 malicious nodes introduced in the network In above simulation 100 nodes (blue “o”) have been deployed in the network. Base Station (green “x”) is deployed within the network. Network is divided in grids. 20 malicious nodes (red “<>”) are introduced in the system which causes some nodes to die (dead node- red “o”). But these malicious nodes are caught and isolated from the system. Scenario 2: Table 6 shows the effect of introduction of 20 malicious nodes in the network terms of first dead and last dead.
• 9.
  International Journal onComputational Science & Applications (IJCSA) Vol.5, No.4, August 2015 95 Table 2. Detection and isolation of malicious nodes [ Figure 5. number of round in which first dead node occur when simulation is performed in the network that does not contain any malicious node and the network that contains malicious node but has been detected and isolated by the proposed protocol Figure 6 . Number of round in which last dead node occur when simulation is performed in the network that does not contain any malicious node and the network that contains malicious node but has been detected and isolated by the proposed protocol Simulatio n Run Network in which no malicious node introduced When 20 malicious nodes have been introduced Number of round when first node occurs Number of round when last node occurs Number of round when first node occurs Number of round when last node occurs Run1 662 2075 619 1982 Run2 732 1994 665 1981 Run3 748 1998 731 1958 Run4 739 2013 708 1997 Run5 648 2014 598 1942
• 10.
  International Journal onComputational Science & Applications (IJCSA) Vol.5, No.4, August 2015 96 Above simulation results have used been obtained by using ZBCA. Table 6, Figure 15 and Figure 16 shows that when no malicious node was introduced, the node died at minimum of 648 round n umber corresponding to which last dead node occur at 2014 and maximum of 748 round number corresponding to which 1998 is the number of round at which last dead node occur. When malicious node is introduced in the system, the first dead occurs at minimum of 619 corresponding to which last dead node occurs at 1982 and maximum of 731 round number to get first dead with 1958 as corresponding last dead round number. It has been observed that the introduction of malicious nodes decrease s network life time i.e. nodes start dying at early stages. But our security algorithm enables the nodes that are affected by such malicious nodes, to detect them very early thus minimizing the corruption of sensed information. Also there is not much difference created in the network lifetime in terms of first and last dead even after introduction of malicious nodes. 12.2.3 Scenario 3: Comparison between networks that use clustering algorithm and those that use no clustering algorithm has been performed as shown by following table: Table 3. Comparison of clustering and non-clustering architectures Figure 7. Number of round in which first dead node occur when simulation is performed in the network that does not use any Clustering architecture and the network that use a Clustering architecture Simulation Run Network in which no clustering and no grid architecture is used Network in which only clustering is used Numbe r of round when first node occurs Number of round when last node occurs Number of round when first node occurs Number of round when last node occurs Run1 474 4058 722 951 Run2 342 3874 668 999 Run3 349 4076 724 989 Run4 426 3983 721 990 Run5 484 3950 686 942
• 11.
  International Journal onComputational Science & Applications (IJCSA) Vol.5, No.4, August 2015 97 Figure 8. Number of round in which last dead node occur when simulation is performed in the network that does not use any clustering architecture and the network that uses a clustering architecture Above simulation result show that when no clustering was used, first dead occur at minimum of 34 2 round number corresponding to which 3847 is the round number in which last dead node occurs. In this architecture first dead node occur at maximum of 474 and corresponding to this 4058 is the round number in which last dead occur. So, the network is highly unstable because first dead node appeared very early and last dead node very late. This happens because in this particular case nodes directly communicate with BS. So if a node was deployed near to BS, it dies late as comparison to node that is deployed at far distance from BS. With use of clustering, first dead node occur at minimum of 668 round number with corresponding 999 round number for last dead n ode and maximum of 724 round number for first dead corresponding to which 989 is the round number for last dead. So network becomes stable. Although last dead node is appearing earlier in clustered network but first dead node appears late. So on average network life time has improved. Scenario 4: Comparison between grid and non grid architecture has been observed. In grid architecture JOIN request was send to all neighboring z one nodes by CH to select the proper c luster head whereas in non grid architecture no such request was send. Nodes send their data to the nearest CH available. Energy consumption in both cases is given as follow: a) Energy consumption for cluster head selection in grid architecture ETX= Energy of data trasmission ERX= Energy of data reception EDA= Energy of data aggregation E consume= Energy to send JOIN request to all node in network+Energy for the reception of data for all of its own and neighboring zone nodes in cluster+Send the CH information to Base station. = (ETX* ctrPacketLength * Distance from selected CH and all other node in the network)+(ERX*ctrPacketLength *N) + EDA+(ETX * ctrPacketLength + Emp* ctrPacketLength) b)Energy consumption for cluster head selection in non grid architecture E consume = Energy for the reception of data for all nodes in cluster+Send the CH information to Base Station.
• 12.
  International Journal onComputational Science & Applications (IJCSA) Vol.5, No.4, August 2015 98 =(ERX*ctrPacketLength *N) + EDA + (ETX * ctrPacketLength + Emp* ctrPacketLength) Table 4. Comparison between grid and non-grid architecture Figure 9. Number of round in which first dead node occur when simulation is performed in the Non- Gird and Grid network Simul ation Run Network in which only clustering is used Network in which both grid and clustering is used Number of round when first node occurs Number of round when last node occurs Number of round when first node occurs Number of round when last node occurs Run1 722 951 1045 2043 Run2 668 999 1002 2040 Run3 724 989 943 2011 Run4 721 990 1016 2041 Run5 686 942 960 2034
• 13.
  International Journal onComputational Science & Applications (IJCSA) Vol.5, No.4, August 2015 99 Figure 10. Number of round in which last dead node occur when simulation is performed in the Non- Gird and Grid network Above results show that in non-grid architecture, first dead and last dead nodes appear earlier in comparison to the case where grid architecture were used. First dead node in non-grid architecture on average is found at 704 which improved to an average of 993 number of round for grid architecture. In terms of last dead node, 974 is the n umber of round in non-grid architecture which increased to 2033 for grid architecture. So network life time in grid architecture is better than any other cases because grid architecture balances the energy consumptions for all sensor nodes in a network field. Network lifetime is improved in terms of first dead and last dead node in grid architecture. Thus an improvement has been made in WSN using ZBCA. Better network lifetime is observed and energy is efficiently used 9.CONCLUSIONS Denial of sleep attack is a great threat to lifetime of sensor networks as it prevents the nodes from going into sleep mode. We have shown the effect of malicious environment on heterogeneous and homogeneous WSN. The basic idea is that when malicious event is introduced, working of sensors gets interrupted because of which the sensors are not able to perform their work of data collection properly and die in very early stages of network life. This means that a node has been compromised or is out of function which compromises the security of data also. So, the critical aspect of applications in WSN is network lifetime. Comparison has been made between Clustering and Non-Clustering architecture. The proposed protocol provides security against Sleep Deprivation Attack. This proposed protocol has been made only for homogeneous network but can be extended for heterogeneous environment. Also we assumed that Cluster Heads will never become malicious node. Improvements can be made by providing security for the scenario where even Cluster Head can be attacked and captured by the attacker. We sincerely hope that our work will contribute in providing further research directions in the area of security.
• 14.
  International Journal onComputational Science & Applications (IJCSA) Vol.5, No.4, August 2015 100 REFERENCES [1] Bhattasali T., Chaki R.: “A Survey of Recent Intrusion Detection Systems in Wireless Sensor Network”, Advances in Network Security and Applications: Conference Proceedings of Fourth International Conference on Network Security and Applications (CNSA 2011), Chennai, July 15-17, 2011, ISBN:978-3-642-22539-0. [2] Brownfield M. , Gupta Y., Davis N. ,“Wireless Sensor Network Denial of Sleep Attack”, Proceedings of the 2005 IEEE, Workshop on Information Assurance, United States Military Academy, West Point, NY, June 2005,. [3] C. Karlof and D. Wagner, “Secure Routing in Wireless Sensor Networks: Attacks and Countermeasures”, Journal of Ad Hoc Networks, Elsevier, 2013. [4] Chen C., Hui L., Pei Q., Ning L., Qingquan P. , “An Effective Scheme for Defending Denial-of- SleepAttack in Wireless Sensor Networks”, Proceedings of the 2009 Fifth International Conference on Information Assurance and Security, Volume 2002, ISBN: 978-0-7695-3744-3. [5] David R.Raymond and Scott F.Midkiff Virginia tech “Denial of service in wireless sensor networks; attacks and defenses”, published by IEEE 2008. [6] Deng, J., Han, R., & Mishra, S., “ A Performance Evaluation of intrusion-Tolerant Routing in Wireless Sensor Networks”, Information Processing in Sensor Networks, Volume 2634, 2003. [7] Deng, J., Han, R., & Mishra, S. (2004). Countermeasures against traffic analysis in wireless sensor networks”, 1st IEEE/CerateNet Conference on Security and Privacy in Communication Networks, September 2005. [8] D. Wood and J. A. Stankovic, “Denial of service in sensor networks,” IEEE Computer, Volume 35, Issue 10, 2002. [9] Eschenauer, L., & Gligor, V. D., “A key-management scheme for distributed sensor networks”, ACM Conference on Computer and Communications Security, November 2002. [10] Ghosh A.K., Bairagi A.K., Kashem M. A., Reszwan-ul-Islam, Uddin A.J., “Energy Efficient Zone Division Multihop Hierarchical Clustering Algorithm for Load Balancing in Wireless Sensor Network”, International Journal of Advanced Computer Science and Applications, Volume 2, Issue 12, December 2011. [11] Idris M.Atakli, Hongbing Hu, Yu Chen,Wei-Shinn Ku and Zhou Su, “Malicious node detection in Wireless Sensor Networks using Weighted Trusted Evaluation”, SpringSim, 2008. [12] I.F. Akyildiz, W. Su, Y. Sankarasubramanian, E. Cayirci, “Wireless sensor networks: a survey”, Computer Networks, Volume 38, 2002.