Exploiting and Handling Delayed Measurements in Sensor Systems

Distributed sensor networks have found numerous real life applications. However the endemic ambiguity of real environments can disrupt the communication procedure such that measurements arrive late to an active agent that estimates the dynamics of the measured phenomenon. In this work we evaluate the informational utility of delayed packets, and show through simulation their impact for filtering algorithm performance. This evaluation demonstrates that routing protocols need to not only optimize packet rate related performance but also minimize estimation error. Using a collaborative signal processing scenario we explicitly compare delayed and undelayed information, by using a-priori estimations of individual measurement's utility. Without using additional communication steps we can achieve significant performance improvement compared to simply discarding delayed measurements.