Signal Detection in an Impulsive Noise Environment Using Locally Optimum Detection

Locally optimum detection is a technique for reliable signal estimation in the presence of strong non-Gaussian interference. It is particularly suited to direct-spread spectrum systems due to the diversity gain achieved because of spreading. However, even more gains can possibly be derived by performing this operation in an iterative fashion, thereby allowing signal detection at even lower signal-to-interference ratios. In this paper the locally optimum detector used in an iterative scheme to suppress strong non-Gaussian interference is studied. The interference is modeled as Middleton class-A and the data stream is convolutionally encoded. Simulation results demonstrating performance improvement over a simple linear combiner are presented.