AI-Native End-to-End Network Slicing for Next-Generation Mission-Critical Services

Abstract

Radio access networks have recently witnessed impressive strides thanks to numerous cutting-edge technologies including network slicing. While these improvements are expected to continue, optimal end-to-end network slicing requires an accurate abstraction of the core network. To properly meet the challenges associated with next-generation networks and services, research and standard organizations envision a revolutionary redesign from the ground up where artificial intelligence will no longer simply be an overlaid service but rather be the foundation upon which all core network functions natively run, i.e., AI-Native. In this first-of-a-kind work, we optimize end-to-end network slicing while considering the 3GPP core network functions and workloads by solving a holistic mixed-integer nonlinear programming problem to minimize the end-to-end latency. Due to its complexity, we decompose it into two sequential problems: a convex access-end problem and an integer linear programming core network-end problem, the latter of which is solved by AI-Native at the core network. Finally, we discuss our extensive simulation results to validate our approach.