Multi-Objective Optimization for Energy-Efficient and Reliable Transmission in WBANs: Session-Specific Design Using MODRL

In wireless body area networks (WBANs), ensuring energy efficiency while maintaining reliable transmission poses a significant and inherently conflicting design challenge. This paper presents a multi-objective optimization (MOO) design for wireless transmission in the finite blocklength (FBL) regime within a WBAN environment, aimed at minimizing both energy consumption and packet error rate (PER) for each transmission session. We propose to train an intelligent agent that can determine near-optimal transmission parameter values for different user preferences. Specifically, we reformulate the MOO problem as a multi-objective Markov decision process (MOMDP) and develop a multi-objective deep reinforcement learning (MODRL)-based solution to approximate the Pareto front, where the penalty-based boundary intersection (PBI) approach is utilized for reward scalarization. Simulation results demonstrate that our proposed solution can adapt transmission parameter values for each transmission session to balance energy efficiency and reliability requirements based on user preferences. Moreover, compared to the linear scalarization method, our solution with the PBI scalarization approach provides a more complete Pareto front.