Towards Cross-Layer Optimization of Virtualized Radio Access Networks

摘要

This paper proposes an approach to cross-layer optimization of virtualized Radio Access Network resources in future mobile communications. Assuming that the Virtual Network Operators (VNOs) guarantee contracted Service Level Agreements (SLAs) with the users, the proposed approach uses weighted proportional fairness as a basis for allocation of the remaining capacity. This allocation is achieved by a distributed, pricing-based solution to a two-layer convex optimization problem. Through this mechanism, some of the key functionalities of the centralized virtualization platform are transferred to the individual VNOs and users. This allows for a drastic reduction in the complexity of the system management compared to the previously proposed centralized approaches. Therefore, it leads to a much more scalable design for dense network deployments with real-time applications. Another advantage of the proposed distributed cross-layer optimization is the enhanced level of isolation among different VNOs. The proposed approach is evaluated by simulating a scenario with 3 types of VNOs and differentiated SLAs sharing radio resources from an underlying physical heterogeneous network. Results for the 4 types of service classes confirm that given sufficient aggregate capacity, all SLAs are satisfied, the entire aggregated capacity is utilized, and the residual available capacity is shared among the users proportionally fair.