Architectures For Real-Time Sequential Detection

We present the sequential detection for diffusion type signals both in the fixed probability of error formulation and in the Bayesian formulation. The optimal strategy in both cases is a threshold policy with explicitly computable thresholds. We provide numerical schemes for approximating the revelant likelihood ratio and provide an architecture for real time signal processing.