Learning-Based Resource Allocation for MBRLLC and Homogeneous Slices in 6G Networks

Network slicing is one of the important components that allows 5G/6G networks from levering the performance with respect to its predecessors. Indeed, network slicing provides the ability to divide the network virtually into slices where each slice is constructed to serve a certain purpose and constraints regarding for instance delay and data rate. Most prior work in this area considered three types of services, namely enhanced mobile broadband (eMBB), ultra-reliable and low-latency communications (URLLC) and reliable massive machine type communication (mMTC). In this paper, we investigate further including beyond 5G (B5G) and 6G services, such as mobile broadband reliable low latency communications (MBRLLC) that encompass the services with both data rate and delay requirements. We consider the radio access network (RAN) slicing, and study the joint problem of inter-slicing and intra-slicing to allocate the resources to the slices and their users in a fair way, with the objective of maximizing a utility function that consists of a weighted sum of the slices' performance. This problem is subject to many constraints, such as the limited radio resources and the slices' requirements in terms of data rate and delay. We formulate the problem as an integer nonlinear program, and propose a deep Q-learning-based inter-slicing algorithm, and a heuristic intra-slicing algorithm to find an approximate solution. Simulation results show that considering the MBRLLC slice yields better results in terms of satisfied users. Moreover, the proposed scheme consistently outperforms the benchmark based on the uniform intra-slicing algorithm.