Opportunistic network coding for two-way relay fading channels

When designing two-way relay channels, there is a dilemma of how to reduce the transmission power by network coding (XORing symbols of opposite directions) with a low symbol delay. Moreover, if the channels are fading, an extra penalty caused by error probability should be added to each transmission, which makes the decision-making problem more complex. In this paper, we propose a new model that considers instantaneous signal to noise ratio (SNR) in addition to queue occupation status. In this model, the channel state evolution is traced by a finite state Markov chain. We develop an efficient computational solution utilizing value iteration algorithm to find an optimal policy regarding symbol delay, transmission power consumption, symbol loss due to the queue overflow and transmission error probabilities. Simulation results show that there is an improvement in both the symbol loss rate and the overall system cost in practical scenarios, compared to the conventional modeling method where channel states are ignored.