Exploiting Spatial and Temporal Correlations for Signal-Centric MAC in M2M Communications

Abstract

In this paper, we propose a signal-centric medium access control scheme that simultaneously exploits spatial and temporal correlations among sensing results for machine-to-machine communications. To model sensing results with spatial and temporal correlations, we propose using a vector autoregressive process. To minimize the overall prediction error, we propose using the space-time predictive polling algorithm. In addition, we derive the a-priori prediction error that is essential for making an optimal polling decision. Simulation results show that the proposed space-time predictive polling algorithm could significantly outperform algorithms that use only temporal correlation.