Artificial Intelligence Based Flexible Preamble Allocation for Radio Access Network Slicing in 5G Networks

Abstract

One of the most difficult challenges in Radio Access Network (RAN) slicing occurs in the connection establishment phase where multiple devices use a common Random Access Channel (RACH) to gain access to the network. It is now very well known that RACH congestion is a serious issue in case of sporadic arrival of machine-to-machine (M2M) nodes and may result in significant delay for all nodes. Hence, RACH resources are also needed to be allocated to different services to enable RAN slicing so that the resources can be dynamically allocated.

In the RACH procedure, the nodes transmit a selected preamble from a predefined set of preambles. If multiple nodes transmit the same preamble at the same RACH opportunity, a collision occurs at the eNodeB. In order to isolate one service class from others during this phase, one approach is to allocate different preamble subsets to different service classes. Static allocation of those subsets, however, may result in inefficiencies when the traffic generated by each service changes significantly over time. Hence, dynamic allocation is more suitable to be able to keep the delay and collision probabilities around the desired levels.

This work proposes adaptive preamble subset allocation methods using Deep Reinforcement Learning (DRL) and Genetic Algorithm (GA). The proposed methods can distribute preambles among different service classes according to their priority and the traffic in the network, providing a virtual isolation of service classes. The results indicate that the proposed mechanisms can quickly adapt the preamble allocation according to the changing traffic demands of service classes.