Enabling distributed detection with dependent sensors

Computational issues affecting the feasibility of optimal distributed detection with correlated measurements are well recognized. We propose utilizing the t-cherry junction tree, an approach based on probabilistic graphical models, to approximate the joint distribution of the correlated measurements. In principle, this approach provides a sequence of progressively more efficiently represented approximations that enable tradeoff between fidelity and compactness. Practically, however, the impact of generating estimated distributions from training data can be significant as the number of parameters to estimate in a distribution grows exponentially with the number of random variables in the distribution. This limitation is quantified and the performance of this approach is illustrated via simulation studies.