Managing data latency for consensus over dynamic sensor-networks

In this publication, we propose a computation scheme to compensate for data latency in dynamic networks performing a distributed consensus task. To convey concepts, we apply the proposed methods for localizing a noise source of interest through a distributed sensor network. The solution generalizes past results by dealing with time-varying non-uniform delays with an unknown upper bound. In particular, this publication presents a relatively simple edge-weight adaptation scheme improving consensus for situations where sensors have either synchronized or unsynchronized clocks. Simulations of a noise source localization task are used to demonstrate and quantify benefits in a statistically significant manner. The suggested scheme reduces the mean value of the range error by 10-15%, thus resulting in an improvement in localization performance.