Research On Distributed Filter Algorithm Of Wireless Sensor Network

In recent years, with the fast development of the network communication technology,digital processing technology and computer technology, wireless sensor network has been ahot area of research from all walks of life. Wireless sensor network detects and collects thedata within its visual range，then feeds the information to the people or other equipmentswhich control the entire network. The development of distributed computing technologypromotes the practical application of sensor network; its advantages make target trackingbased on wireless sensor network remarkable achievements and major breakthroughs in otherapplications. However, the sensors are easily disturbed by the outside world. The perceptionof the same sensors in different regions is different from the perception of different kinds ofsensors to the same target. The uncertainty makes the data between the neighbors unreliableor we say the nodes are less trustworthy，which lastly impairs the estimation accuracy of theentire algorithm. How to address the bad impact of the uncertainty as well as the distrustfulenvironment? It is an urgent problem. Therefore this paper gives some countermeasures forimproving the estimation accuracy of the sensor network. This paper summarizes the researchresults, and focuses on the distributed filtering algorithm in the application of the actualnetwork.This paper makes some improvements based on the classical algorithm, which considersa single target tracking and estimation using multiple sensors as a core. Aiming at the problemthat each node has different certainty degree to the target in sensor network withpacket-dropping, an adaptive consensus-based linear distributed filtering algorithm isproposed and the theoretical analysis of the stability of the algorithm is also given. Under thecondition of the same packet-dropping rate, we make the simulation analyses andcomparisons between the improved algorithm and the classical algorithm. Similarly, in thecase of the different packet-dropping rate, we also do the simulations. The results show thatthe proposed algorithm has a more accurate estimate in the packet-dropping case, as well as inan ideal case. Secondly, the interference of the external environment, which means a non-trustenvironment, makes sensor data unreliable and the sensor nodes are vulnerable to be attacked,which lastly affects accuracy of the target tracking and location. To improve the estimationaccuracy for the target tracking of the sensor network, a distributed filtering algorithm basedon tunable weights is proposed, and we give the simulation analyses between the proposedalgorithm and the classical algorithm under the random attack model and selective attackmodel. Lastly, simulation results demonstrate the effectiveness of the proposed algorithm.