Latency Synchronization for Social VR with Mobile Edge Computing

While mobile edge computing (MEC) potentially supports the stringent latency requirements for Virtual Reality (VR), previous research only considers minimizing the latency of transmitting required data and ignores the latency consistency of a group of friends in social VR. In this paper, we leverage the human motion prediction to eliminate the inconsistency among users in a group while ensuring the low latency interaction between friends with the aid of MEC. Moreover, to capture the accurate users’ interaction behaviors, it is critical to jointly predict the motions of users with spatially close proximity in VR. Accordingly, we formulate a new problem, named Group Motion Prediction for Social VR problem (GMSV), with group consistency requirement and the objective of prediction costs minimization on the MEC server. Then, we prove that GMSV is NP-Hard and design a new algorithm, named Social-Aware Latency Synchronization for Remote Users (SLSR), to select an appropriate set of remote users in social VR for motion prediction, with the ideas of generating a partial order set of remote subgroups and extracting the remote subgroups with high social interaction utilities. Simulation results show that SLSR can effectively increase latency consistency by more than 50% compared with the baseline schemes.