Scalable Service Chaining in MEC-Assisted 5G Networks

This paper focuses on a Mobile Edge Cloud (MEC)-assisted 5G network architecture in which an optical transport network is used to interconnect MEC servers processing low layer functions and Central Cloud servers processing upper layer functions of the LTE protocol stack. Traditionally, the problem of Based Band Unit (BBU) function assignment in cloud environments has been addressed through Integer Linear Programming (ILP) formulations having the disadvantage of increased computational complexity. To address this issue, we initially propose a relaxation framework that reduces computational complexity. To further accelerate convergence, the proposed relaxation algorithm is coupled with Hierarchical Random Graph (HRG) theory. Taking advantage of the hierarchical structure of the 5G communication and compute infrastructure, HRG theory can be employed to generate simple tree topologies that reflect the statistical properties of the originally complex 5G infrastructure. Numerical results indicate that the proposed approach outperforms traditional centralized C-RAN schemes with much faster convergence times.