Virtual Reality (VR) technology is being advanced along the lines of enhancing its immersiveness, enabling multiuser Virtual Experiences (VEs), and supporting unconstrained mobility of the users in their VEs, while constraining them within specialized VR setups through Redirected Walking (RDW). For meeting the extreme data-rate and latency requirements of future VR systems, supporting wireless networking infrastructures will operate in millimeter Wave (mmWave) frequencies and leverage highly directional communication in both transmission and reception through beamforming and beamsteering. We propose to leverage predictive context-awareness for optimizing transmitter and receiver-side beamforming and beamsteering. In particular, we argue that short-term prediction of users' lateral movements in multiuser VR setups with RDW can be utilized for optimizing transmitter-side beamforming and beamsteering through Line-of-Sight (LoS) "tracking" in the users' directions. At the same time, short-term prediction of orientational movements can be used for receiver-side beamforming for coverage flexibility enhancements. We target two open problems in predicting these two context information instances: i) lateral movement prediction in multiuser VR settings with RDW and ii) generation of synthetic head rotation datasets to be utilized in the training of existing orientational movements predictors. We follow by experimentally showing that Long Short-Term Memory (LSTM) networks feature promising accuracy in predicting lateral movements, as well as that context-awareness stemming from VEs further benefits this accuracy. Second, we show that a TimeGAN-based approach for orientational data generation can generate synthetic samples closely matching the experimentally obtained ones.

翻译：虚拟现实（VR）技术正沿着增强沉浸感、支持多用户虚拟体验（VE）以及通过重定向行走（RDW）将用户约束在专门VR设备的同时实现无约束移动等方向不断发展。为了满足未来VR系统对极端数据速率和低延迟的要求，支撑无线网络基础设施将在毫米波（mmWave）频段运行，并通过波束赋形和波束控制利用高度定向的收发通信。我们提出利用预测性上下文感知来优化发射端和接收端的波束赋形与波束控制。具体而言，我们论证了在多用户VR场景中结合RDW时，对用户横向移动的短期预测可用于通过视距（LoS）“跟踪”用户方向来优化发射端波束赋形与波束控制。同时，对方向性移动的短期预测可用于接收端波束赋形以增强覆盖灵活性。我们针对预测这两类上下文信息中的两个开放性问题展开研究：（i）基于RDW的多用户VR环境中的横向移动预测，以及（ii）生成用于训练现有方向性移动预测器的合成头部旋转数据集。我们通过实验证明，长短期记忆（LSTM）网络在预测横向移动方面具有显著精度，并且源自VE的上下文感知可进一步提升这一精度。其次，我们展示了基于TimeGAN的方向性数据生成方法能够生成与实验数据高度匹配的合成样本。