Channel gain prediction for multi-agent networks in the presence of location uncertainty

Paper in proceeding, 2016

Coordination among mobile agents relies on communication over a wireless channel and can thus be improved by channel prediction. We present a Gaussian process framework to learn channel parameters and predict the channel between arbitrary transmitter and receiver locations. We explicitly incorporate location uncertainty in both learning and prediction phases. Simulation results show that if location uncertainty is not modeled appropriately, it has a degenerative effect on the prediction quality.