Multi-criteria Service Function Chaining Orchestration for Multi-user Virtual Reality Services

Immersive entertainment based on Multi-User Virtual Reality (MUVR) is gaining popularity to enable in-game interactions with multiple users. However, computing and network-intensive utilization of resources require changes in cloud-based and traditional monolithic (single machine) deployments towards distributed edge computing architecture to enable high-quality interaction. Decomposing MUVR into elements with specialized Service Functions (SF) in order and organizing it into a Service Function Chaining (SFC) requests orchestration enables MUVR to reuse frame processing between users and improves MUVR scalability. However, orchestrating distributed edge computing resources and multiple destination SFCs while minimizing delay and meeting resource requirements is a challenging task. This article proposes a multi-criteria SFC orchestration scheme for MUVR services, called MuSFiCO. MuSFiCO maps edge computing resources and instantiates SFCs on distributed servers based on delays threshold, CPU and memory resources, and bandwidth. We developed a constrained-based heuristic to minimize delay and compare it with the baseline monolithic deployment and cloud-based SFC algorithms. Results demonstrate the efficiency of MuSFiCO compared to other approaches in terms of latency, CPU, memory, bandwidth utilization, as well as orchestration decision-time.