Joint route selection and split level management for 5G C-RAN
This work tackles the problem faced by network/infrastructure providers of jointly selecting routing and functional split level to satisfy requests from virtual mobile network operators (vMNOs). We build a novel system model that brings together all the involved elements and features, embracing spli...
| Authors: | , , , , , |
|---|---|
| Format: | article |
| Publication Date: | 2021 |
| Country: | España |
| Institution: | Universidad de Cantabria (UC) |
| Repository: | UCrea Repositorio Abierto de la Universidad de Cantabria |
| Language: | English |
| OAI Identifier: | oai:repositorio.unican.es:10902/23426 |
| Online Access: | http://hdl.handle.net/10902/23426 |
| Access Level: | Open access |
| Keyword: | Route selection Functional split 5G C-RAN |
| Summary: | This work tackles the problem faced by network/infrastructure providers of jointly selecting routing and functional split level to satisfy requests from virtual mobile network operators (vMNOs). We build a novel system model that brings together all the involved elements and features, embracing split levels defined by the 3GPP and packet switch fronthaul network. To our best knowledge, this is the first work that provides a solution for multiple vMNO requests considering the two aforementioned sub-problems (i.e. split selection and routing). We use the model defined to formulate an optimization problem, which is characterized by the exponential size of its search space. We propose two heuristic approaches to address this problem: (1) a greedy scheme, and (2) an evolutionary algorithm, which is also improved with a specialized initialization. We conduct extensive experiments to assess the performance and behavior of the proposed methods, over varying network instances. When possible, we also perform comparisons with respect to the optimal solution and a well-known commercial solver. Our results indicate that the proposed techniques represent appropriate trade-offs between solution quality and execution time, and can serve complementary goals: the quality of the results yielded by our evolutionary method are better, but at the cost of longer execution times; in contrast, our greedy algorithm offers a reasonably appropriate performance, with an execution time that is notably lower. Our experiments show that it is possible to produce near-optimal results to the above complex problem through computationally efficient algorithmic solutions. |
|---|