Latent parameter estimation in fusion networks using separable likelihoods

Multi-sensor state space models underpin fusion applications in networks of sensors. Estimation of latent parameters in these models has the potential to provide highly desirable capabilities such as network self-calibration. Conventional solutions to the problem pose difficulties in scaling with the number of sensors due to the joint multi-sensor filtering involved in the evaluation of the parameter likelihood. We propose an approximation with a node-wise separable structure which can be evaluated by local filtering operations, instead. When leveraged with Markov random field models and message passing algorithms for inference, these likelihoods facilitate scalable estimation across the network and fit in both centralised and decentralised processing paradigms. We relate the approximation quality of the proposed separable likelihoods to the accuracy of state estimation using joint and local filtering schemes. We demonstrate this approach for network self-localisation using measurements from non-cooperative targets in an example.