Energy-aware Embedding of Logical Functionality Chains over Heterogeneous Platforms
Khadijeh Shahsavand, Amina Hentati, Mouhamad Dieye, Roch Glitho · 2024
In the 5G radio access network, baseband functions are divided into the radio unit (RU) and two logical entities, namely the distributed unit (DU) and the centralized unit (CU). In the core network, the User Plane Function (UPF), another logical entity, serves as a singular data plane element. A key factor influencing performance and overall network efficiency is the placement of 5G logical functionalities, including DUs, CUs, and UPFs. While most existing research on provisioning focuses on homogeneous platforms such as virtual machines (VMs), practical implementations often use a variety of platforms, including bare-metal servers and containers alongside VMs. This paper proposes an energy-efficient placement and chaining strategy for 5G logical functionalities across heterogeneous platforms, addressing traffic demands with varying quality-of-service requirements. After presenting a novel formulation of the problem as an integer linear programming (ILP), we introduce a heuristic ranking-based algorithm (HRA) to solve it efficiently. Extensive simulation results demonstrate that our algorithm can reduce power consumption by up to 30% and improve acceptance rate by up to 40% while satisfying both latency and bandwidth constraints. Moreover, our comparative analysis of homogeneous and heterogeneous networks underscores the importance of choosing a virtualization strategy that aligns with the specific requirements of network functionalities to optimize power consumption.