In this paper, we propose signal-centric predictive medium access control algorithms for machine-to-machine communications. In particular, the proposed signal-centric predictive algorithms exploit temporal correlations among sensing results to reduce the total prediction error at the base station. We propose polling algorithms for optimal performance and distributed algorithms for scalability. In addition, we derive analytical results that characterize the polling schedule when the proposed signal-centric predictive polling algorithm is used. Furthermore, we propose algorithms that strike a good balance between prediction error and fairness. Simulation results show that the proposed algorithms significantly reduce the overall prediction error.
- Machine-to-machine communications
- mean square estimation
- medium access control
- statistical signal processing
- stochastic processes