Intelligent Session Management for URLLC in 5G Open Radio Access Network: A Deep Reinforcement Learning Approach

To sustain ultra-reliable and low latency communication for the fifth generation (5G) networks, the latency of data forwarding over the core network is conventionally ignored. To significantly reduce the latency, a base station shall not permit to service a new session before the case of unacceptable latency taking place. To this end, the fundamental challenge turns out to proactively cognize that the requirements of reliability/latency are about to be violated. To address this challenge, in this article, a deep reinforcement learning based intelligent session management for the open radio access network is proposed to efficiently allocate the resources for the serving sessions and new sessions. The experimental testing results sufficiently show the practicability of our scheme for the 5G networks.