Packet Scheduling for Wireless Powered Communication Systems in the Finite Blocklength Regime: A POMDP Approach

This paper proposes a packet scheduling scheme in a wireless powered internet-of-things (WP-IoT) system with the consideration of some recent results on finite blocklength (FBL). The system includes one source node (SN) and one hybrid access point (HAP). The SN is assumed to have no embedded energy supply and thus need to firstly perform energy harvesting (EH) from radio signals of HAP. Before performing EH and data transmitting, we assume that the SN needs to first estimate the channel quality. With the assumption of imperfect channel estimation process, the packet scheduling problem in the proposed system is formulated as a partially observable Markov decision (POMDP) problem. To maximize the system long-term throughput, the optimal policy for packet scheduling is derived by a dynamic programming approach. Simulation results demonstrates that the proposed packet scheduling scheme outperforms the corresponding myopic scheme.