AEIJST - November 2015 - Vol 3 - Issue 11 ISSN - 2348 - 6732 A Comprehensive Study on Data Aggregation Techniques in Wireless Sensor Networks * Dharmendra Parmar ** Prof. Jaimala JHA * (CSE/IT Dept.), MITS, Gwalior (M.P) ** Asst. Prof. CSE/IT Dept., MITS. Gwalior (M.P) Abstract Wireless sensor networks comprises of sensor nodes. These networks have enormous application in environmental observation, calamity management, security and defences, etc. Wireless sensor nodes are extremely smaller and contain restricted processing ability extremely minimum battery power. The limitation of battery power makes the sensor network vulnerable to failure. Data aggregation a very vital method in wireless sensor networks. With the aid of data aggregation decrease the energy utilization by evading redundancy. In this paper, we discussed about data aggregation and its various energyefficient techniques used for data aggregation in WSN as well as presenting diverse sorts of architectures, its requirements, classification at last its advantages and disadvantages. Keywords - WSN; Data aggregation; classification; architectures I. Introduction The Wireless Sensor Network (WSN) is composed by computational devices known as sensor nodes. Each node has a processing, storage, sensing and communication units. A WSN may have thousands of sensor nodes, and their deployment might be planned or random. Considering that network topology is unpredictable before node deployment, WSN requires self-organization ability, similar to traditional ad hoc networks [1]. Wireless Sensor Networks collect data from large geographic areas and can be utilized for several applications (i.e. environment monitoring, space exploration, military operations). Energy consumption is a concern, as sensor nodes have restricted energy capacity and they could be deployed on inhospitable areas. Thus, sensor nodes are designed to be disposable when their energy supply is exhausted. Random deployment imposes management challenges. Sensor nodes should react to environment characteristics such as noise and interference. In addition to that, environment and network changes require sensor nodes a continuous process of self-adaptation and self-reconfiguration. These functions are a subset of self-management services for WSN [2]. One of the characteristics that distinguish WSN from other networks is its focus on data. Instead of providing communicating services and resources sharing, the main goal of WSN is to collect and provide data. This requires researches to develop new and specific protocols and architectures for WSN that must also consider that the network can have high density and that energy is a constraint. The amount of sensor nodes deployed over a monitoring area may reach the order of hundreds or thousands. The architecture must be able to efficiently communicate wirelessly, and can´t spend too much energy, since a WSN may require lasting for years. II. Data Aggregation Data aggregation is the procedure of gathering and aggregating the useful information. Data aggregation is regarded as one of the essential processing method for conserving the energy. In WSN, data aggregation is an efficient manner to conserve the restricted resources. The foremost objective of data aggregation algorithms is to collect and aggregate data in an energy proficient way so that network lifetime is prolonged [3]. Page 1 of 10 www.aeph.in AEIJST - November 2015 - Vol 3 - Issue 11 ISSN - 2348 - 6732 The data aggregation is an approach utilized to resolve the implosion and overlap issues in data-centric routing. Data arriving from several sensor nodes is combined as if they are about the similar attribute of the phenomenon when they reach the same routing node on the way back to the sink. Data aggregation is a broadly employed approach in wireless sensor networks. The security issues are data confidentiality and integrity. In data aggregation become essential when the sensor network is equipped in hostile surroundings. Data aggregation is a procedure to aggregate the sensor data by using aggregation approaches [4]. The general data aggregation algorithm operates as shown in the below fig1. Fig.1 demonstrates that data aggregation is the procedure of aggregating the sensor data using aggregation approaches. Then the algorithm employs the sensor data from the sensor nodes and then combines the data by employing certain aggregation algorithms namely centralized approach, LEACH (low energy adaptive clustering hierarchy), TAG (Tiny Aggregation) etc. This aggregated data is transferred to the sink node by electing the efficient path. Data collected from sensor nodes Data aggregation algorithms Aggregated data Base Station Fig. 1 Process of data aggregation A. Requirement of Data Aggregation Sensor nodes are deployed in remote environments to a multi-hop WSN [3] over a broad area. Extremely not often do the users have global information on the sensor node‟s distribution. Consequently, when users request state-based sensor readings of the attributes namely heat and moisture in a random area, networks may suffer the irregular heavy traffic. This issue requires data aggregation to observe user necessities and handle overlapped aggregation trees of numerous users proficiently. Numerous useful applications like ecological observing, military applications, and so on, are investigating the utilization of WSNs. Such applications need transferring a vast amount of relevance, observed data from one side of the network to other. Since WSNs are mostly deployed with low power batteries, battery life is a key constraint in any real-time application. This requires the utilization of energy competent data dissemination protocols for aggregation of the observed data. Nodes of a WSN in close proximity generally hold related data due to a property called spatial correlation [3]. Page 2 of 10 www.aeph.in AEIJST - November 2015 - Vol 3 - Issue 11 ISSN - 2348 - 6732 III. Data Aggregation Based Networks Broadly classified into two classes [5] A. Flat Networks Flat networks play very vital part in wireless sensor network, in which every sensor node have an equivalent battery power and plays the similar kind of role in a network. In such type of networks, data aggregation has to be done in data centric routing manner, where the sink generally sends a data packet to the sensor nodes, such as, flooding. In the flooding sensors, which have data identical, transmit reply data packet back to the sink. B. Hierarchical Networks All the communication and computation load at the sink in flat network, that‟s why lot of energy is utilized. In the hierarchical network, in which data aggregation data has to be done at special nodes, with the aid of this special node can decrease the amount of data packet forwarded to the sink. So with this network enhances the energy competence of the entire network [5]. IV. Architectures of Data Aggregation The Based on various applications and necessities there are numerous existing architectures for data aggregation [6]. A. Centralized Architecture Centralized architecture is extremely simplest architecture of wireless sensor network. Where data fusion procedures is applied. Every sensor nodes observes a data and send to the one central node, called central processor fusion node, this node fuse the reports gathered by all sensor nodes. In this architecture central node have a reliability of whole network. The basic benefits of this architecture are it can be straightforwardly detect wrong report of information which is taken by the completely wireless sensor network. The shortcoming is that nonflexible to sensor changes and the workload is concerned at a single point. Sensor Node Cluster Head Fig. 2 Centralized architecture B. Decentralized Architecture The decentralized architecture of wireless sensor network, there is no single centralized node that makes decisions on behalf of all the sensor nodes. Data fusion takes place locally at every node based on local observations and the information received from neighbour nodes. In which all sensor nodes are linked to each other on the observation. The benefits of this architecture are scalable and tolerant to the addition or loss of sensing nodes or dynamic changes in the network. Page 3 of 10 www.aeph.in AEIJST - November 2015 - Vol 3 - Issue 11 ISSN - 2348 - 6732 Sensor node Fig.3 Decentralized Architecture C. Cluster Based Architecture Wireless sensor network is resource constraint that‟s why sensor cannot straightforwardly send data to the base station. In which all regular sensors can transmit data packet to a head of cluster, which collect data packet from all the regular sensors in its cluster and transmit the brief digest to the base station. With the aid of the approach conserves the energy of the sensors. In energy-constrained sensor networks of large size, it is ineffective for sensors to send the data straightforwardly to the sink. In such situations, sensors can broadcast data to head of cluster, which collects data from all the sensors in its cluster and transmits the short digest to the sink. There are certain problems involved with the process of clustering in a wireless sensor network. First issue is, what number of clusters ought to be created that could optimize certain performance parameter. Second could be number of nodes must be taken in to an individual cluster. Third imperative issue is the election process of cluster-head in a cluster. sink Cluster Head Sensor Node Fig.4. cluster based architecture D. Tree Based Architecture In the tree-based scheme, execute aggregation by building an aggregation tree, which could be a minimum spanning tree, rooted at sink and source nodes are regarded as leaves. Every node has a parent node to send its data. Flow of data begins from leaves nodes up to the sink and therein the aggregation done by parent nodes. In which every node are managed in form of tree means hierarchical, with the aid of intermediary node we can execute data aggregation procedure and data send leaf node root node. One of the key features of tree-based networks is the building of an energy efficient data-aggregation tree. Data aggregation Source Node Data aggregation Source Node Fig.5. Tree based Page 4 of 10 www.aeph.in AEIJST - November 2015 - Vol 3 - Issue 11 ISSN - 2348 - 6732 E. Chain Based Architecture In this architecture, each sensor transmits data to the nearer neighbor. All sensors are ordered into a linear chain for data aggregation. The nodes can create a chain by using a greedy algorithm or the sink can choose the chain in a centralized way. In the Greedy, chain formation assumes that all sensors have comprehensive knowledge of the network. The furthermost node from the sink starts chain formation and, at every step, the nearest neighbor of a node is elected as its successor in the chain. In every data-collecting round, a node gets data packet from one of its neighbors, collects the data with its own, and transmits the collected data packet to its other neighbor along the chain. Ultimately, the leader node is related to head of cluster transmits the collected data to the base station. Leader Sensor node Sink node Fig.6. Chain based F. Grid Based Architecture In which a set of sensors is assigned as data aggregators in fixed area of the sensor network. The sensors in a grid transmit the data packet straightforwardly to the aggregator of that grid. Hence, the sensors within a grid do not communicate with each other. In-network aggregation is comparable to grid-based data aggregation with two foremost differences; each sensor within a grid communicates with its neighboring node. Any node within a grid can assume the role of aggregator node in terms of rounds until the last node dies. This is identical to cluster-based data aggregation in which the head of clusters are fixed. In in-network aggregation, the sensor with the most crucial information collects the data packets and transmits the fused data to the sink. Every sensor sends its signal strength to its neighbors. If the neighbour has higher signal strength, the sender stops sending packets. After receiving data packets from all the neighbours, the node that has the highest signal strength becomes the data aggregator. The in-network aggregation approach is best appropriate for environments where an event is extremely localized. Fig.7 Grid based V. Related Work 1.Shu Qin Ren et al [7], proposed an efficient and robust data aggregation scheme for Cognitive Wireless Sensor Network, which has a good resilience against node compromise, malicious disruption and intermediate eavesdropping attacks by integrating an end-to-end encryption and a group based density mining method in each cluster for data aggregation. First, the unique subgroup keys improve the scheme resilience against node compromise. One compromised node will not reveal the other group nodes‟ sensitive information. Page 5 of 10 www.aeph.in AEIJST - November 2015 - Vol 3 - Issue 11 ISSN - 2348 - 6732 Second, the end-to-end encryption algorithm can preserve data privacy from forwarding node, aggregator nodes and even to the base station. Even the aggregator can extract data similarity from each small group of concealed data; it cannot dig out the whole data distribution. At last, the scheme make use of secure comparison on concealed data within subgroups, which aids the aggregator to filter out the scarce or malicious data from compromised node or manipulating environment. 2. Wang et al in [8] devised a density-awareness and delay-sensitive data collecting approach for wireless sensor networks. This approach mutually uses static sink and mobile sink to gather data from the entire network. In such manner, this approach can resolve the premature network partition issue, resulting from the huge traffic load in the static sink's area. On the other hand, the coexistence of static sink assures the delay of data delivery for source nodes cannot wait to transmit their data until the arrival of mobile sink. Contrasting with previous algorithms, the foremost assistance of density-awareness and delay-sensitive relies in, mobile sink can adapt the time to live of fresh position announcement message as per to the related node density, such that performance can be enhanced considerably in terms of packet delay and packet delivery ratio with no sacrifice of energy utilization balance in the entire network, other if a source node only has the path to static sink, packets originated from it may be redirected to the closer mobile sink when pass through immediate node. Finally, they results demonstrate that densityawareness and delay-sensitive outperforms the previous data collecting approach in terms of packet delay, data delivery ratio and network lifetime. 3. Kasirajan et al in [9], presented a unique adaptive compression approach utilizing nonlinear estimation theory for data aggregation. Adequate performance of the presented compression approach in the existence of noise, deformation, and quantization errors is showed employing Lyapunov scheme. The presented approach is compared with previous compression approaches exploiting several metrics applicable to wireless sensor networks aggregation can enhance the over-all energy conserving with a small level of deformation which relay on the solution of the Quantizer and the number of aggregation levels rather than network size. 4. Yeganeh et al in [10] presented a structure-free data aggregation procedure, for gathering delay-constrained data in WSNs. To attain this, it merges a dynamic Real-time Data-aware Routing policy and a Judiciously Waiting policy. The first scheme considers data types as well as real-time decisions to elect next hop neighbour nodes with better aggregation performance and increments spatial convergence during transmissions without explicit maintenance of a structure while the second one presents artificial delays and increases temporal convergence. The presented approach is extremely appropriate for saving energy in real-time WSNs. 5. Khan et al in [11] addressed the problem of energy-efficiency in WSNs in the perspective of network self-organization and clustering. Author devised a new and competent network structure for an energy-efficient deployment of WSNs. Firstly presented a new Zone-Based Hierarchical Framework to arrange the network, followed by a new Zone-Based SelfOrganization Clustering approach for energy efficient clustering. The main uniqueness in this work relies in proposing a new zone-based structure, which segments the network into n-zones and assigning every zone with a zone manager node. Additionally, the structure utilizes an n-tier hierarchal management scheme of managing and managed node. The presented Zone-Based Hierarchical Framework and Zone-Based SelfOrganization self-organize the network into clusters and considerably decrease the energy utilization during the network self-organization clustering process. 6.Bhoi et al in [12] devised a scheme of detecting the fault by employing Density-Based Clustering scheme. The primary concept is to create a density-based cluster in which the nodes within the cluster have similar behavior. Page 6 of 10 www.aeph.in AEIJST - November 2015 - Vol 3 - Issue 11 ISSN - 2348 - 6732 The cluster is created by utilizing ε-Neighborhood, in which the Density-Reachability and Density-Connectivity concepts are exploited to get the Density-Based Cluster. By this scheme, the faults are identified as the nodes, which are not in the cluster. Initially, find a point and according to the ε-neighborhood of the point we form a circle. „ε‟ shows the radius of the circle with respect to the node. If the ε-neighborhood of a node or point contains minimum number of points then this node is called core object. So, like this we find the points or nodes which are of same behavior (according to the assumptions taken) and then come to know which node is faulty and which node is not faulty. 7. Jing He et al in [13] addressed the essential issues of constructing a load-balanced data aggregation tree in probabilistic WSNs. Author concentrated on constructing a LoadBalanced Data Aggregation Tree underneath the PNM. More particularly, three issues are examined, namely, the Load-Balanced Maximal Independent Set issue, the Connected Maximal Independent Set issue, and the load-balanced data aggregation tree construction issue. Load-Balanced Maximal Independent Set and Connected Maximal Independent Set are well-known NP-hard problems and load-balanced data aggregation tree is an NPcomplete problem. Subsequently approximation algorithms and comprehensive theoretical examination of the approximation aspects are presented in the paper. 8. Kim et al in [14] devised a density control scheme based on disjoint wakeup scheduling which can give a full availability to sink node with a minimum set of active nodes in an exceedingly dense system to enhance the life span of the network. Author devised a disjoint scheduling algorithm for density control in wireless sensor network. It chooses a minimum set of working nodes to evade wasting of energy in excess by turning off too many duplicates nodes. It chooses a set of minimum active nodes to enhance energy efficiency while providing path accessibility to sink node. 9. Khanjary et al in [15] introduced aligned-orientation directional sensor networks in which nodes are equipped relaying on Poisson point procedure and the orientation of all sensor nodes is the same. Then, devised a scheme to estimate density of nodes at decisive percolation for both of the sensing-coverage phase transition (SCPT) and network connectivity phase transition (NCPT) issue in such networks, for all angles of field-of-view between 0 and π by employing continuum percolation. Due to percolation theory, the decisive density is infimum density that for densities above it sensing-coverage phase transition and network connectivity phase transition almost certainly take place. Additionally, proposed a model for percolation in directional sensor networks, which provides a basis for resolving the sensing-coverage phase transition and network connectivity phase transition issue collectively. 10. Gayathri et al in [16] presented an efficient data gathering scheme in respect to lower the utilization of energy, data loss and end-to-end delay in wireless sensor network by presenting reference point imparting mechanism (RPIM). The sensor nodes function though batteries and it is difficult to recharge/replace repeatedly. Energy conserving is always critical to the presence of the networks due of restricted sensing coverage and network connectivity. This work aims to reduce the frequent energy drain with the sensor nodes for enhancing the life of the network. In reference point imparting mechanism, the reference points are selected relayed on their connectivity and it ensue the data from every sensor node. The mobile collector is utilized to save energy within sensor nodes for collecting the data. Its path is definite by imparting method and obtains the summarized data from every reference point. Hence, the energy required for data transmission using reference point imparting mechanism is compact and energy is utilized proficiently. 11. Hassan Harb et al in [17] studied a novel prefix-suffix filtering method for data aggregation in periodic sensor networks. Further, explore the issue of discovering all pair of nodes creating comparable data sets. Append a new suffix frequency filter method to the previous prefix frequency filtering. Page 7 of 10 www.aeph.in AEIJST - November 2015 - Vol 3 - Issue 11 ISSN - 2348 - 6732 The objective is to combine additional filtering scheme in respect to lessen the latency of the aggregation stage. The key idea behind this technique is to make the prefix filtering condition more difficult to be satisfied by introducing a new suffix filtering condition based on the frequency order. The objective of our technique is to reduce the number of redundant data sent to the end user while preserving the data integrity. We developed a new suffix frequency filter technique beside the existing prefix frequency filtering technique. We use this additional filtering technique that prunes erroneous candidates that survive after applying the prefix and frequency filtering technique. 12. Bagaa et al in [18]: This paper investigated the data aggregation principle and focused on features and requirements that should be implemented by data aggregation scheduling protocols. In particular, the paper reviewed existing solutions by proposing a clear taxonomy to classify these works. The surveyed solutions aim to reduce the data latency, increase the data accuracy and aggregation freshness with a good distribution of nodes‟ waiting time (WT). According to this one, we have classified existing solutions into two classes: the first one is un-slotted data aggregation scheduling and works at routing layer; whereas the second one is slotted data aggregation scheduling and relies on a cross layer design (Routing and MAC). Unlike the un-slotted-based algorithms, which consider only the time of data aggregation process when distributing WT over network nodes, slotted based solutions consider the time of both data aggregation process and communications. For this reason, in recent years the slotted-based solutions attracted more attention than the un-slotted-based ones. Finally, we shed some light on new directions and open issues. 13. Zhao et al in [19] devised and examined an energy efficient Compressive sensing (CS) based approach, which concurrently realizes efficient data aggregation and clustered routing in wireless sensor networks called “Treelet-based Clustered Compressive Data Aggregation” (T-CCDA). Particularly, as a first step, Treelet transform is adopted as a sparsification tool to mine sparsity from signals for Compressive sensing recovery. The scheme not only improves the performance of CS recovery, but also reveals localized correlation structures within sensor nodes. Then, a unique clustered routing algorithm is devised to further make possible energy conserving by taking advantage of the correlation structures. 14. Rezvani et al in [20], introduced a novel collusion attack scenario against a number of existing IF algorithms. Moreover, author devised an enhancement for the IF algorithms by providing an early estimate of the trustworthiness of sensor nodes which makes the algorithms not only collusion robust, except also more exact and quicker advancing. VI. Advantages of Data Aggregation Data aggregation offers numerous advantages. [6] With the aid of data aggregation process, we can improve the robustness and correctness of information, which is obtained from whole network. Certain redundancy exists in the data gathered from sensor nodes thus data fusion processing is required to decrease the redundant information Other benefits is those lessens the traffic load and savs energy of the sensors VII. Disadvantages of Data Aggregation Data aggregation procedure has its own disadvantages some are listed below [6] The head of cluster implies data aggregated nodes forwards aggregated data to the base station. This head of cluster may possibly be attacked by suspicious attacker. If a head of cluster is compromised, then the base station (sink) cannot be ensuring the accuracy of the aggregated data that has been forwarded to it. Page 8 of 10 www.aeph.in AEIJST - November 2015 - Vol 3 - Issue 11 ISSN - 2348 - 6732 Other shortcoming in previous systems is multiple replica of the aggregated outcome may be transmitted to the base station (sink) by uncompromised nodes. It increase the power utilized at these nodes. Conclusion Data aggregation is the method of gathering and aggregating the useful data. Data aggregation is regarded as one of the essential processing measures for conserving the energy. In WSN, data aggregation is an efficient means to conserve the restricted resources. The foremost aim of data aggregation algorithm is to collect and aggregate data in an energy proficient way so that network lifetime is prolonged. In this paper, we have provided a brief introduction regarding data aggregation, its type, architectures, the requirement of data aggregation and its advantages and disadvantages. References: [1] J´ulio C´esar e Melo, Linnyer Beatrys Ruiz;” Data-Centric Density Control for Wireless Sensor Networks”. The Fourth International Conference on Wireless and Mobile Communications IEEE, 2008 [2] L. B. Ruiz, J. M. Nogueira, and A. A. F. Loureiro. Manna: A management architecture for wireless sensor networks. IEEE Communications Magazine, 41(2):116–125, 2003. [3] Ankit Tripathi, Sanjeev Gupta, Bharti Chourasiya;” Survey on Data Aggregation Techniques for Wireless Sensor Networks”. International Journal of Advanced Research in Computer and Communication Engineering Vol. 3, Issue 7,July 2014 [4] Mousam Dagar and Shilpa Mahajan;” Data Aggregation in Wireless Sensor Network: A Survey”. International Journal of Information and Computation Technology.,2013 [5] Sushruta Mishra, Hiren Thakkar “Features of WSN and Data Aggregation techniques in WSN: A Survey”. International Journal of Engineering and Innovative Technology(IJEIT) Volume 1, Issue 4 ,April 2012 [6] Kiran Maraiya, Kamal Kant, Nitin Gupta “Architectural Based Data Aggregation Techniques in Wireless Sensor Network: A Comparative Study”. International Journal on Computer Science and Engineering 2011 [7] Shu Qin Ren, Jong Sou Park;” Density Mining Based Resilient Data Aggregation forWireless Sensor Network”. Fourth International Conference on Networked Computing and Advanced Information Management, IEEE, 2008 [8] WANG Jie-tai, XU Jia-dong, Liang Hua-qiang;” A Density-awareness and Delaysensitive Data Collecting Scheme for Wireless Sensor Networks”. IEEE, 2009 [9] Priya Kasirajan, Carl Larsen and S. Jagannathan;” A New Adaptive Compression Scheme for Data Aggregation in Wireless Sensor Networks”. IEEE, 2010 [10] Mohammad Hossein Yeganeh, Hamed Yousefi, Naser Alinaghipour, Ali Movaghar;” RDAG: A Structure-free Real-time Data Aggregation Protocol for Wireless Sensor Networks”. 17th IEEE International Conference on Embedded and Real-Time Computing Systems and Applications. IEEE, 2011 [11] Muhammad Zahid Khan, Madjid Merabti, Bob Askwith, FaycaJ Bouhafs;” A ZoneBased Hierarchical Framework and Clustering Scheme for Energy-Efficient Wireless Sensor Networks”. IEEE, 2012 [12] Sourav Kumar Bhoi, Sanjaya Kumar Panda, and Pabitra Mohan Khilar;” A DensityBased Clustering Paradigm to Detect Faults in Wireless Sensor Network”. Springer, 2013 Page 9 of 10 www.aeph.in AEIJST - November 2015 - Vol 3 - Issue 11 ISSN - 2348 - 6732 [13] Jing (Selena) He, Shouling Ji, Yi Pan, Yingshu. Li;” Constructing Load-Balanced Data Aggregation Trees in Probabilistic Wireless Sensor Networks”. IEEE, 2013 [14] EunHwa Kim;” A Density Control Scheme Based on Disjoint Wakeup Scheduling in Wireless Sensor Network”. Springer, 2014 [15] Mohammad Khanjary, Masoud Sabaei, Mohammad Reza Meybodi;” Critical Density for Coverage and Connectivity in Two-Dimensional Aligned-Orientation Directional Sensor Networks Using Continuum Percolation”. IEEE Sensors Journal, Vol. 14, No. 8, August 2014 [16] Gayathri Deyi S ,NandhaKumar R YaraLakshmi P:” A New Energy Consumption Technique In Wireless Sensor Network Using Reference Point And Imparting Mechanism”. International Conference on Electronics and Communication System (JCECS -2014), 2014 [17] Hassan Harb and Abdallah Makhoul, Rami Tawil and Ali Jaber;” A Suffix-Based Enhanced Technique for Data Aggregation in Periodic Sensor Networks”. IEEE, 2014 [18] Miloud Bagaa, Yacine Challal, Adlen Ksentini, Abdelouahid Derhab, and Nadjib Badache;” Data Aggregation Scheduling Algorithms in Wireless Sensor Networks: Solutions and Challenges”. IEEE Communications Surveys & Tutorials IEEE, 2014 [19] Cheng Zhao, Wuxiong Zhang, Yang Yang and Sha Yao;” Treelet-Based Clustered Compressive Data Aggregation for Wireless Sensor Networks”. IEEE Transactions on Vehicular Technology IEEE, 2014 [20] Mohsen Rezvani, Aleksandar Ignjatovic, Elisa Bertino, and Sanjay Jha;” Secure Data Aggregation Technique for Wireless Sensor Networks in the Presence of Collusion Attacks”. IEEE Transactions on Dependable And Secure Computing (TDSC) IEEE, 2014 Page 10 of 10 www.aeph.in