The Arbitrarily Varying Channel With Colored Gaussian Noise

We address the arbitrarily varying channel (AVC) with colored Gaussian noise. The work consists of three parts. First, we study general discrete AVC fixed parameters, a model that combines and time-varying channel. determine both deterministic code capacity random capacity. Super-additivity is demonstrated, showing can be strictly larger than weighted sum parametric capacities. In second part, consider product (AVGPC). Hughes Narayan characterized through min-max optimization leading to “double” water filling solution. As in case standard AVC, discontinuous input constraint, depends on which or state constraint higher. opposed Shannon's classic solution, it observed coding using independent scalar codes suboptimal for AVGPC. Finally, establish noise, where double performed frequency domain. analysis relies our preceding results, parameters