GROVE: A Cost-Efficient Green Radio over Ethernet Architecture for Next Generation Radio Access Network

TL;DR

This paper introduces GROVE, a cost-efficient, environmentally friendly Ethernet-based architecture for next-generation radio networks that combines function splitting, renewable energy, and optimized routing to reduce costs and improve scalability.

Contribution

The paper proposes a novel integrated model, GROVE, combining function splitting, renewable energy, and Ethernet routing for C-RAN, formulated as a MILP for cost minimization.

Findings

Joint optimization of routing, energy, and function splitting reduces costs.

GROVE outperforms classical disjoint approaches in all tested scenarios.

Scalability analysis identifies limits of the MILP solver for larger networks.

Abstract

Centralized/Cloud Radio Access Network (C-RAN) comes into prominence to reduce the rising energy consumptions and maintenance difficulties of the next-generation networks. However, C-RAN has strict delay requirements, and it needs large fronthaul bandwidth. Function splitting and Radio over Ethernet are two promising approaches to reduce these drawbacks of the C-RAN architecture. Meanwhile, the usage of renewable energy sources (RESs) in a C-RAN boosts the energy-efficiency potential of this network. In this paper, we propose a novel model, which is called Green Radio OVer Ethernet (GROVE), that merges these three approaches to maximize the benefits of C-RAN while maintaining the economic feasibility of this architecture. We briefly explain this model and formulate an operational expenditure minimization problem by considering the several restrictions due to the network design and the service provisioning. Then we linearize the quadratic routing decision constraints in the problem to solve it with a mixed-integer linear programming (MILP) solver. Results show that it is cost-effective to choose routing, function splitting, and RES decisions together. Our solution surpasses classical disjoint approaches for all studied cases. Besides, we provide a network scalability analysis to determine the MILP solver's limits for larger network topologies.