A Deep Reinforcement Learning Approach for Automated Chamber Configuration Replicating mmWave Directional Industrial Channel Behavior

Industrial wireless channels have different characteristics than home and office channels due to their reflective nature. Moreover, the millimeter-wave (mmWave) wireless bands can play a big role in improving industrial wireless systems due to their large available bandwidth and the short wavelength that allows a large number of antennas to be located closely to each other. Wireless test chambers are used for over-the-air (OTA) testing and assessment of various protocols and equipment. However, in order to closely characterize a system under test, the chamber should be configured to replicate the environment where the system is deployed. In this work, we present a deep reinforcement learning protocol to configure a test chamber in order to replicate the spatial characteristics of measured mmWave channels in industrial environments. The proposed algorithm is general for any N-dimensional chamber configurations where it can be used to configure various reflectors, absorbers, and paddles inside a wireless test chamber.