I. INTRODUCTION
With the vigorous development of fifth-generation mobile communication technology and beyond (5G/B5G), various innovation services such as industrial automation, unmanned vehicles, virtual reality (VR), remote training, etc., are constantly emerging [1], [2]. However, the stringent low-latency requirement has become one of the toughest challenges for the implementation of these emerging services [1]. In particular, after the advent of ultra-high speed intermediate nodes and 100 Gbps wired links, the 5G access network (5G-AN), as a solution in the "last mile" of data delivery, has become a bottleneck that determines the end-to-end (E2E) delay of the entire network [3].