Low-Power Distributed Kalman Filter for Wireless Sensor Networks

Abstract

Distributed estimation algorithms have attracted a lot of attention in the past few years, particularly in the framework of Wireless Sensor Network (WSN). Distributed Kalman Filter (DKF) is one of the most fundamental distributed estimation algorithms for scalable wireless sensor fusion. Most DKF methods proposed in the literature rely on consensus filters algorithm. The convergence rate of such distributed consensus algorithms typically depends on the network topology. This paper proposes a low-power DKF. The proposed DKF is based on a fast polynomial filter. The idea is to apply a polynomial filter to the network matrix that will shape its spectrumin order to increase the convergence rate by minimizing its second largest eigenvalue. Fast convergence can contribute to significant energy saving. In order to implement the DKF in WSN, more power saving is needed. Since multiplication is the atomic operation of Kalman filter, so saving power at the multiplication level can significantly impact the energy consumption of the DKF. This paper also proposes a novel light-weight and low-power multiplication algorithm. The proposed algorithm aims to decrease the number of instruction cycles, save power, and reduce the memory storage without increasing the code complexity or sacrificing accuracy.