Q-learning Based Random Access with Collision free RACH Interactions for Cellular M2M

Abstract

This paper investigates the coexistence of M2M and H2H based traffic sharing the RACH of an existing cellular network. Q-learning is applied to control the RACH access of the M2M devices which enables collision free access amongst the M2M user group. Frame ALOHA for a Q-learning RACH access (FA-QL-RACH) is proposed to realise a collision free RACH access between the H2H and M2M user groups. The scheme introduces a separate frame for H2H and M2M to use in the RACH access. Simulation results show that applying Q-learning to realise the proposed FA-QL-RACH scheme resolves the RACH overload problem and improves the RACH-throughput. Finally the improved RACH-throughput performance indicates that the FA-QL-RACH scheme has eliminated the collision between the H2H and M2M user groups.