Performance analysis of a power-constrained cross-layer scheduling using artificial neural networks

Cross-layer scheduling has received an intense attention recently owing to the desire to exploit protocol interactions to design optimum systems that are adaptable to service requirements of network applications. This article discusses a Lyapunov stable power-constrained opportunistic scheduler that makes an optimum use of the wireless spectrum while guaranteeing a minimum service to all flows active in multi-flow, multiuser wireless systems. We formulate the constrained scheduling problem as a dynamic system of differential equations. We then establish a Lyapunov function for the dynamic system associated with the search for an optimum solution for the constrained convex optimization problem, and then apply Lyapunov stability theory to prove the system's convergence. The system of differential equations is used to excite a neural network whose outputs are the solutions to the constrained optimization problem.