Quickest change detection in distributed sensor systems

Abstract- In the conventional formulation ofthe changepoint detection problem, there is a sequence ofobservations whose distribution changes at some unknown point in time, and the goal is to detect this change as quickly as possible, subject to false alarm constraints. It is known that in the case where the observations are independent and identically distributed (iid) and the change point is modeled as a geometrically distributed random variable, the Shiryaev detection procedure minimizes the expected detection delay, subject to a constraint on thefalse alarmpmbabilify In this paper; we present effective decentralized detection procedures for the multi-sensor situation where the information available for decision-making is distributed across a sef of sensors. We present asymptotically optimal procedures for two scenarios. In the prst scenario, fhe sensors send quantized versions of their observations to afusion center where the change detection is pedormed based on all the sensor messages. In the second scenario, the sensors perform local change detection using ShiryaPv-Roberrs procedures and send theirfinal decisions to thefusion centerfor combining. We show that our decentralized pmcedures for latter scenario have the samefirst order asymptotic performance as the centralized Shiryaev-Roberts procedure that has access to all of the sensor observations. We also present numerical results for a simple example involving Gaussian observations.