Networked estimation using compressed data
Christian Musso, K. Benameur · 2017
In this paper, we consider different approaches in reducing the amount of data transfer in a distributed Kalman filtering based on noisy linear observations. The observations are either compressed using equivalent measurements, or transmitted only if their values change more than a specified value. The objective is to reduce sensor data traffic with relatively small estimation performance degradation. Through simulations, we evaluate the performances of a distributed Kalman filter using three techniques for data transfer.