Adversaries resilient distributed parameter estimation algorithm in Wireless Sensor Networks

Abstract

This thesis studies adversaries resilient distributed parameter estimation of an unknown target parameter when some adversarial nodes are existed in Wireless Sensor Networks(WSNs). Proposed algorithm is consensus + innovation estimator in which the nodes of networks update their estimate combining their previous estimate, neighborhoods’ estimates and local measurement. And the algorithm detects the adversarial nodes using time-varying attack threshold value. When attacks are detected, the algorithm raises the flag to alert the attack and update the parameter that is used in consensus part in estimation using the difference between node’s estimate and its neighborhood’s estimate. Finally, we demonstrate the performance of proposed algorithm with several attack examples.