A time series of decisions approach in detection systems

Abstract

Time series of decisions in detection systems enable us to integrate the individual sensor's decisions to improve the detection performance. This paper considers the problem of optimum fusion structure in time series of decisions detection systems. The optimum local fusion structure that minimizes the total risk is derived. It is found that the optimum local fusion structure is the majority logic fusion strategy. The proposed structure is applied in the case of a single sensor detection system with Gaussian and exponentially distributed observations. Monte Carlo simulation is provided to evaluate the system performance. Significant performance improvement is achieved using the proposed structure. This structure enables us to exploit the benefits of data fusion even when we use a single sensor detection system.