Learning-Based Low-Latency VIoT Video Streaming Against Jamming and Interference

Abstract

Video streaming of the Visual Internet of Things (VIoT) has to guarantee the quality of service for latency-sensitive multimedia applications against jamming and interference. In this article, we propose a reinforcement learning (RL)-based low-latency VIoT video streaming scheme, which enables base stations to choose the streaming policy without relying on the known channel and jamming model. In this scheme, the video compression encoding rate, the modulation and coding rate, the transmit channel, and the power level are chosen according to the received signal strength indicator, the buffer queuing length, the received jamming power, and the previous video streaming quality. A deep RL version is proposed for the base stations with sufficient computational resources, in which three deep neural networks are designed to provide accurate policy distribution and Q value approximations, and the policy entropy is evaluated to avoid choosing the local optimal policy. We discuss their computational complexity and formulate the game model between the jammer and the base station. Simulation results are provided to validate the efficacy of our proposed schemes.