MIQA: An Application Agent for Immersive Content Delivery Over Millimeter Waves

The highly directional nature of the millimeter wave (mmWave) beams causes several challenges in using that spectrum to meet the communication demands of immersive applications. The mmWave beams are especially susceptible to misalignments and blockages caused by user movements. As a result, mmWave channels are vulnerable to large quality fluctuations, which in turn, degrades the end-to-end performance of immersive applications. In this paper, we propose a reinforcement learning (RL) based application-layer plugin that works in conjunction with the QUIC protocol to combat the challenges of mmWave networks. The plug-in called Millimeter wave based Immersive QUIC Agent (MIQA) uses the RL model to help modulate the sending rate along with the congestion control scheme of QUIC. To evaluate the effectiveness of MIQA, we conduct experiments on a mmWave augmented immersive testbed. The evaluation results show that MIQA significantly improves the immersive experience by increasing the end-to-end throughput and by decreasing the end-to-end latency.