Opportunistic Scheduling For A Two-Way Relay Network Using Markov Decision Process

In this study, the authors study transmission scheduling for a two-way relay network in time-varying fading channels, where the relay node can opportunistically use traditional one-way relay technique or network coding to forward traffic to the end nodes. They formulate a stochastic dynamic programme with the objective of minimising the long-run cost, defined as a function of both the transmission power and data transmission delay. An unconstrained Markov decision process model is developed and solved for the average and discounted cost problems. The optimal solution requires high computational and modelling complexity when the state space is large. For this reason, they develop heuristic solutions with lower complexity. For the discounted cost problem, a simulation-based dynamic programming algorithm is proposed that not only simplifies the modelling process and reduces the computational complexity, but also achieves close-to-optimum cost. For the average cost problem, a heuristic scheduling scheme is proposed, which makes transmission decisions based on estimated costs in the current and next time slots. The heuristic scheme achieves close-to-optimum cost performance while greatly reducing the computational complexity.