Rule-Sensitive Sensor Participation in Distributed Detection with Prescribed Boolean Fusion Rules

Abstract

We study distributed binary detection when the fusion node must implement a prescribed Boolean decision rule. Under conditionally independent Bernoulli observations, we derive a signed-influence path identity valid for arbitrary Boolean rules; for monotone rules, it reduces to an exact Russo–Margulis representation in terms of coordinate influences. The identity yields rule-sensitive necessary conditions for a target Bayes error and an influence-weighted Kullback–Leibler feasibility bound. We then connect Boolean relevance to exact rule implementation and derive lower bounds on the number of sensors that must participate, including a heterogeneous bound based on the ordered per-sensor hypothesis separations. Numerical and Monte Carlo results illustrate both the tightening provided by the heterogeneous participation bound and the performance penalty that can arise when a prescribed fusion rule is retained instead of optimal likelihood-ratio fusion.