Distributed Estimation And Security Mechanisms For Wireless Sensor Networks

With features of low-power, low-cost, easy to deploy and self-organization, wireless sensor networks(WSNs) have been widely used monitoring of sensitive information in diverse fields such as military, environment, household and industrial. But a wireless sensor network is vulnerable since it is deployed in harsh environments. In addition, WSNs suffer from a lot of constraints, including small memory, low computation capability and limit energy source. Thus, secure wireless sensor network is facing great challenges.Attacks in wireless sensor networks include the following common situation. The attacker may disrupt routing protocols to reduce network performance, or consume a large of network energy to make the network paralysis, or inject false data to change thefinal result. In this thesis, we deal with the secure problems of distributed consensus of WSNs and we study the security of the monitored environment for WSNs. We focus on the following three challenging works.1. A secure distributed consensus scheme for wireless sensor networks. Considering data falsification attack, especially covert false data injection attacks, we present a distributed detection mechanism in which sensor node compare deviation with the threshold to detect abnormal nodes is presented. And a weighted averaging consensus scheme is proposed. Simulation results illustrate the effective of the proposed secure scheme.2. A distributed consensus-based secure scheme for heterogeneous sensor networks. For large-scale field environment, we consider a heterogenous WSNs composed of two types of sensors: sensor nodes and relay nodes. A asymptotically unbiased consensus scheme with adaptive local consensus parameters is proposed. The algorithm allows lower precision nodes to attend the consensus while improves the security of the network.3. Indoor intrusion detection and security for mobile wireless sensor networks.Mobile sensor networks can complete dangerous task instead of human beings. There are obstacles in indoor environment, the speed of sensor nodes is limited. By using the concepts of graph theory, the indoor environment dispersed into finite graph. We establish an upper bound of sensor nodes that can capture the intruders. A hunting strategy based on tree decomposition is proposed to ensure the safety of the indoor environment.