A Study of Pilot Placement Optimization with Constrained MDPs in IEEE802.11p Systems

This paper proposes a decision-assisted pilot placement optimization method in IEEE802.11p physical layer. The fast time-varying channel is first modeled as a typical Gaussian-Markov process. Under the constraint of state-spaces, the pilot optimization problem is further formulated as constrained Markov decision processes (MDPs). Secondly, for achieving compatibility with existing standards, our goal is to determine the optimal pilot placement and employ only very limited pilot patterns in response to fast varying channels. We develop a channel state matched pilot optimization method, where the optimization procedures focus on how to respond the different channel variations in the time and frequency domains. To jointly evaluate the severity of channel variations in the time and frequency domains, we derive an effective mutual information measurement criterion. Simulation and numerical results show the efficiency of the pilot optimization decision scheme in reducing the channel estimation error, and mutual information measurement can yield an accurate performance evaluation in relatively fast time-varying vehicular communication scenarios.