Isolation Aware CU-DU Mapping for Multi-Tenant 5G O-RAN Slices

Resource efficiency is critical for 5G and beyond 5G (B5G) mobile networks. Leveraging different Radio Access Network (RAN) functional splits, baseband functions in 5G are disaggregated into three main components- Radio Unit (RU), Distributed Unit (DU), and Centralized Unit (CU). These disaggregated components can be placed in different geographical locations, leading to a higher flexibility and efficiency in RAN. However, such disaggregation makes the placement of baseband functions challenging due to the constraints imposed by the requirements of network slices, baseband functions, limited capacity in processing nodes and transport links, etc. It becomes even more challenging when slices of multiple tenants have different isolation requirements regarding the baseband functions. In this work, we address the problem of resource-efficient baseband function placement for multi-tenant slices in 5G Open RAN (O-RAN). We formulate the problem as an Integer Linear Programming (ILP) based optimization model to minimize the cost of processing and bandwidth resources. We consider the various requirements of delay, data rate, and sharing policies of multi-tenant slices, as well as limited resource capacity in the network while placing the functions. We perform extensive simulations to analyze the behavior of our model and show that it incurs lesser cost than baselines while placing the functions in the network. To deal with the high computational complexity of ILP, we also propose a low-complexity heuristic algorithm to achieve reasonable performance in significantly less time.