State and Computing Decoupled Service Function Chaining for Scalable Cloud RANs
Zecheng Fang, Jianing Cao, Chunjing Yuan, Lin Tian · 2024
The stateful nature of Network Functions (NF) poses significant challenges for elastic scaling, which is a key promise of Network Function Virtualization (NFV). The concept of state and computing decoupling has been proposed to address the challenge of elastic scaling for stateful NFs, while inevitably impacts the processing performance of NFs due to the remote state access operation. The protocols and functionalities of Radio Access Networks (RANs) are rooted in stateful design. As the evolution of RAN embraces NFV concepts, state placement and message routing, as well as virtual network function (VNF) orchestration, need to be jointly considered to reduce the state access impacts on performance-critical stateful RAN NFs. In this paper, we consider a Cloud RAN running as a Service Function Chain (SFC) and formulate the joint state and VNF orchestration model as a Mixed-Integer Quadratically Constrained Programming (MIQCP) problem to minimize network costs. Furthermore, an efficient Joint State and VNF Orchestration (JS- VNFO) algorithm is proposed to embed SFC requests, while ensuring the latency requirement of mobile users. The performance evaluation shows that the proposed algorithm can achieve lower network costs and significant end-to-end latency reduction compared to the benchmark strategy.