Markov Decision Process Based Pilot Pattern Optimization for 5G V2X Communications

This paper proposes an Markov decision process (MDP) based pilot placement optimization approach for the radio access in 5G vehicle to everything (V2X) communications. The optimal placement problem of pilot symbols is based on a typical pilot-assisted OFDM transmission and simplified to a finite state-space representation. We propose and formulate a finite MDP so as to determine an appropriate pilot pattern from a set of candidate pilot configurations. Additionally, an enhanced pilot placement scheme is developed to reduce the complexity of solving MDP problems. We derive analytical expressions of the mutual information, which to some extent allow us to jointly evaluate the dynamics of the channel state in time and frequency domain. Numerical results show that the proposed pilot optimization policy is capable of improving the channel estimation, and the mutual information based measurement criteria can yield more accurate evaluations in fast time-varying vehicular channels.