Network-aware retransmission strategy selection in ad hoc wireless networks

Retransmission strategies have been widely investigated in wireless networks, since they are able to grant considerable benefits in dynamic environments. Distributed schemes, based on cooperative techniques, can also add the benefits of spatial diversity, particularly if combined with Multi-User Detection decoding schemes. However, the impact of each scheme on the rest of the network cannot be neglected, since it also affects the overall network performance. A mathematical approach to evaluate this impact can be very involved, given the potentially very large number of parameters to take into account. In this paper, we propose instead a probabilistic approach, based on Bayesian Networks, to determine the expected impact, in terms of interference, of different schemes on the rest of the network. Through our framework, it is possible to adaptively select the best scheme to use, as a function of the observation of some topological parameters. We also design a distributed protocol to implement a variety of retransmission schemes, and the performance results confirm the effectiveness of our model over a static choice of the retransmission strategy and also over a selfish retransmission scheme that always selects the strategy that maximizes the probability of success of the retransmission.