Distributed Bernoulli Filtering Using Likelihood Consensus

We consider the detection and tracking of a target in a decentralized sensor network. The presence of the target is uncertain, and the sensor measurements are affected by clutter and missed detections. The state-evolution model and the measurement model may be nonlinear and non-Gaussian. For this practically relevant scenario, we propose a particle-based distributed Bernoulli filter (BF) that provides to each sensor approximations of the Bayes-optimal estimates of the target presence probability and the target state. The proposed method uses all the measurements in the network while requiring only local intersensor communication. This is enabled by an extension of the likelihood consensus method that reaches consensus on the likelihood function under both the target presence and target absence hypotheses. We also propose an adaptive pruning of the likelihood expansion coefficients that yields a significant reduction of intersensor communication. Finally, we present a new variant of the likelihood consensus method that is suited to networks containing star-connected sensor groups. Simulation results show that in challenging scenarios, including a heterogeneous sensor network with significant noise and clutter, the performance of the proposed distributed BF approaches that of the optimal centralized multisensor BF. We also demonstrate that the proposed distributed BF outperforms a state-of-the-art distributed BF at the expense of a higher amount of intersensor communication.