Deep Reinforcement Learning-Driven Wireless Sensor Placement for IIoT-Integrated Smart Production

Sensor placement problems usually use heuristic algorithms or traditional mathematical optimization methods. Although such methods can search for a helpful solution within an acceptable feasible time, the quality of the solution is not guaranteed. Moreover, such methods are mainly designed for static optimization problems, and need to be completely optimized from scratch when disturbances occur. In this paper, a deep reinforcement learning sequential decision is proposed to solve the sensor layout problem under geometric constraints in the field of industrial Internet of Things, which is expressed as a Markov decision process. Furthermore, deep reinforcement learning algorithm is applied to conduct layout planning experiments in wireless sensor layout environment of different tasks, thus verifying the feasibility of deep reinforcement learning method in solving the challenges of wireless sensor layout with geometric constraints. At the same time, for the uncertainty of complex environment in IIOT integrated intelligent production, a method is proposed to design the wireless sensor placement environment and adjust the training logic to enhance its adaptability to solve the specific sensor placement problem discussed in this paper.