Target tracking in asynchronous sensor networks under temporal misalignment

In this paper, we present a framework for fusion under temporal misalignment in asynchronous sensor networks, focusing on cases where the temporal misalignment is not precisely known and has to be estimated. Temporal misalignment arises when sensor readings are not synchronized in time, posing a significant challenge for accurate data fusion. We introduce a fusion approach based on the logarithmic opinion pool where the time offset among sensors is estimated by maximizing the a posteriori probability after fusion. We further discuss how to apply the proposed approach in recursive estimation settings considering both centralized and distributed architectures. Our framework accommodates both single-target and multi-target tracking scenarios, leveraging the Poisson and independent and identically distributed cluster (IIDC) multi-target densities. In all the considered scenarios, we derive closed-form expressions for formulating the joint fusion and time offset estimation problem. Furthermore, we propose Gaussian mixture implementations for efficient computation of fusion factors. Simulation experiments concerning single-target and multi-target tracking problems are presented to demonstrate the effectiveness of the proposed framework.