Research On Privacy-preserving Data Aggregation Algorithm In Wireless Sensor Networks

In Wireless sensor networks, each node sensing, collecting and transmittinginformation about the perceived objects in coverage areas periodically throughcollaboration with neighbors. Each sensor node is always resources limited.Eliminating the redundant information in data-gathering process can reduce theamount of data traffic across the network effectively, reducing the burden ofcommunication and energy consumption of each node,and can extend the network’slifetime and improve the efficiency and accuracy of data gathering through by thetechnology of data aggregation. Generally, a basic data aggregation algorithm doesnot provide privacy-preserving mechanisms. But, a privacy-preserving mechanism isalways essential in some cases such as the application of Health Monitoring, militaryreconnaissance and Private Households.According to the implementation strategy about researches at home and abroad,privacy-preserving mechanisms can be divided into three classes. One class is calledhop-by-hop encryption mechanism, which can deal with internal and external attackseffectively, but the frequent encrypting and decrypting operations lead to highcomputational cost and time delays. The second class is the end-to-end encryptionmechanism. In this mechanism, the direct operations to the cipher information byaggregation nodes is saving time overhead costing by encryption and decryption, andbringing a short time delay. But Homomorphic encryption supports just only a fewoperations. The third mechanism is using those methods without encryption and doesnot require the process of key distribution, not require any operations of encrypting ordecrypting. Thus, the overhead of communication and computing of this approach isthe least in three mechanisms, but their privacy ability depends on the protectiontechnologies used.Meanwhile, node failures caused by environmental factors or the depleted battery willaffect the performance of the network seriously. Also, failure nodes will interfere with adjacent nodes, and even lead to large cavity in coverage areas, affecting the normaloperation of the entire network, reducing the robustness, integrity, and the accuracy ofdata fusion results.It is important to identify failure nodes rapidly and accurately, thatcan help us replenish sensor nodes in cavity areas timely. So we can do better workabout network management and performance testing in an environment whosetopology changes dynamically.Considering the privacy, the confidentiality, the accuracy of data aggregation resultsand other design goals, we proposed a Privacy-Preserving Data AggregationAlgorithm with the Ability of Failure Node Detection abbreviated PPND, which usesthe characteristics of the data aggregation tree structure and a data perturbation aboutthe raw data by adding a pre-allocation random to achieve privacy protection.Meanwhile, PPND can find all failure nodes through running aggregating process tworounds. The first round is used to calculate the children-set of each node under theaggregation tree structure built before and another round is used to find all theunreached nodes that also called failure nodes. Theoretical analysis shows that it isvery difficult for malicious attackers trying to steal node privacy in PPND.Meanwhile, comparing with algorithms of TAG and SMART, simulation results alsodemonstrate a good performance on communications and accuracy of PPND.