Bayesian prediction for dynamic radio interference

This paper proposes a Bayesian prediction framework for the dynamic interference, caused by the movement of the interfering device. The framework relies on a hidden Markov model where the hidden state is the device mobility and the observable is the measured interference. To efficiently predict the interference, we consider jointly exploiting the mobility correlation and the radio correlation in a probabilistic manner. The benefits in doing so is to keep all available information about the estimation meanwhile to have the flexible to handle the algorithm output. The numerical evaluations shows that the proposed approach outperforms the conventional algorithm such as the Auto-Regressive prediction.