Joint Optimization via Deep Reinforcement Learning in Wireless Networked Controlled Systems

This paper proposes a deep Reinforcement Learning (RL) based co-design approach for joint-optimization of wireless networked control systems (WNCS) where the can help achieve optimal performance under network uncertainties e.g. delay and variable throughput. Compared to traditional modern methods dynamics system are important predicting system’s future response, model-free adapt many applications stochastic behaviour. Our work provides comparison how is affected by such as delays bandwidth consumption an unknown number devices. The data transmitted different conditions several transmit background traffic using same network. problem contains sub-optimization problems because devices non-deterministic channel capacity constraints. proposed seeks minimize error in order improve Quality Service Control. used compared three RL Q-learning algorithms high-throughput flow double emulsion droplets formation application. results show that allowable reliable communication bounded constraints 10 when binary search. without considering effect reward function (Scenario 1) was good with C51 algorithm; including OMNet++ 2), best achieved all (C51, DQN, DDQN) exponential function, only case linear function. Finally, random 3), DDQN performed well, but DQN did not converge. Comparisons other machine learning non-machine also highlight superior utilized algorithms.