New algorithm tackles VR streaming bottleneck that kills user experience

This paper investigates the problem of joint uplink and downlink bitrate optimization for multi-user VR $360^\circ$ video streaming with the objective of maximizing overall Quality of Experience (QoE). We address the challenges posed by the high bandwidth demands, latency sensitivity, and strong coupling between uplink feedback and downlink video delivery in immersive VR applications. The optimization problem is formulated as a nonlinear and nonconvex task that explicitly captures key QoE factors, including delivered video quality, playback stalls, and inter-segment quality variations, while also accounting for heterogeneous user conditions and network constraints. Due to the NP-hard nature of the resulting formulation, we introduce a tractable solution approach that relaxes binary decision variables and applies an iterative optimization framework to efficiently approximate the optimal allocation. The proposed method simultaneously promotes fairness among users and enforces smooth quality transitions over time to mitigate cybersickness and user discomfort. We validate the effectiveness of the approach through extensive simulations in realistic multi-user VR $360^\circ $ streaming scenarios with varying bandwidth availability and user population sizes. The results demonstrate substantial QoE gains compared to a representative baseline strategy, alongside improved fairness in bandwidth allocation and stable performance across multiple trials. Furthermore, we extend the framework with a reinforcement learning-based component that enables adaptive, realtime decision making, making the solution suitable for dynamic and live VR streaming environments.