Soft-Decision-Based Distributed Detection With Correlated Sensing Channels

Prior works on distributed detection with correlated Gaussian observations primarily dealt with single-bit reports to the fusion center and suboptimal fusion rules. Considering multiple bits and independent reporting channel errors, we formulate the optimal detection problem as a nonlinear integer programming problem for which the genetic algorithm is applied to find a suboptimal solution. The Bayesian probability of error performance of our suboptimal solution, first, for one-bit, is better than the performance of some previously proposed schemes and, second, for multiple bits, approaches that of the centralized detection scheme.