Abstract
We study the important and challenging problem of virtualized radio access network (vRAN) design in its most general form. We develop an optimization framework that decides the number and deployment locations of central/cloud units (CUs); which distributed units (DUs) each of them will serve; the functional split that each BS will implement; and the network paths for routing the traffic to CUs and the network core. Our design criterion is to minimize the operator's expenditures while serving the expected traffic. To this end, we combine a linearization technique with a cutting-planes method in order to expedite the exact solution of the formulated problem. We evaluate our framework using real operational networks and system measurements, and follow an exhaustive parameter-sensitivity analysis. We find that the benefits when departing from single-CU deployments can be as high as 30% for our networks, but these gains diminish with the further addition of CUs. Our work sheds light on the vRAN design from a new angle, highlights the importance of deploying multiple CUs, and offers a rigorous framework for optimizing the costs of Multi-CUs vRAN.
Introduction: The Complexity of vRAN Design
In the evolution toward 5G and beyond, the Cloud Radio Access Network (C-RAN) architecture is fundamentally changing how cell towers are built. Traditional monolithic base stations are being replaced by a split architecture: Distributed Units (DUs) at the edge handle radio-frequency tasks, while Central/Cloud Units (CUs) in centralized servers pool processing functions.
The 3GPP standard introduces various "functional splits," which define the distribution of processing layers between DUs and CUs. While centralizing more functions offers economies of scale and reduced computing costs, it imposes significant demands on fronthaul networks for high-capacity, low-latency routing. Designing such a network involves a massive combinatorial challenge: determining the optimal number and location of CUs, assigning DUs to these CUs, selecting the appropriate functional split for each base station, and routing traffic efficiently. This paper addresses these dimensions jointly to minimize total operational expenditures.
Methodology: An Exact Optimization Framework
The authors formulate the vRAN design problem as a Mixed-Integer Quadratically Constrained Program (MIQCP). The objective is to minimize total operational expenditures—including computing, VM instantiation, and routing costs—while adhering to strict 3GPP delay and capacity constraints.
To overcome the computational complexity of MIQCP for large-scale networks, a two-stage exact solution framework is proposed:
- Linearization: Intricate quadratic constraints, arising from the interaction of binary variables for assignment and splitting, are replaced with linear auxiliary variables, transforming the problem into a Mixed-Integer Linear Program (MILP).
- Benders' Decomposition: A cutting-planes method splits the MILP into a Master problem (handling discrete decisions like CU placement and functional splits) and a Slave problem (handling continuous routing decisions).
This iterative approach ensures convergence to a provably exact optimal solution, avoiding the cost penalties associated with heuristic methods.
Experimental Evaluation
The framework was tested using two real-world operational RAN topologies:
- N1: Comprising 198 DUs and 15 potential CU locations.
- N2: Comprising 197 DUs and 15 potential CU locations. System parameters, such as traffic loads (150 Mbps), CPU capacities, and path delays, were derived from actual testbed measurements and 3GPP specifications.
Key Insights and Results
The study reveals several critical findings for network operators:
- The Threshold Effect: Strategic deployment of multiple CUs significantly reduces costs. Moving from a single CU to multiple locations yielded savings of up to 16.36% in N1 and 28.79% in N2. However, these benefits diminish once the optimal number of locations is reached; adding more candidate sites beyond this threshold provides no further gain.
- Distributed vs. Centralized Capacity: Deploying multiple CUs with lower individual capacities is more cost-effective (by up to 26.9%) than a single massive CU with equivalent aggregate capacity.
- The Necessity of Optimization: Randomly placing CUs, even in the correct number, is inefficient. The proposed optimization framework outperformed random deployment strategies by 18.87% in total cost savings.
- Resilience to Routing Costs: As routing costs increase, the system naturally favors edge processing. However, optimized multi-CU placement allows the network to maintain high centralization degrees (up to ~77%) even under expensive routing conditions.
Conclusion
This work provides a rigorous framework for optimizing multi-CU vRANs, demonstrating that joint optimization of placement, splitting, and routing is essential for cost-efficient 5G infrastructure. While results are topology-dependent, the framework's generality allows it to be adapted for various operator-specific constraints and cost models. Future research may explore non-linear cost functions and varying levels of required centralization.
Summary updated