Dynamic ACB Scheme Based on Neural Networks and Markov Chain

Abstract

This paper presents an access class barring (ACB) scheme based on neural networks and Markov chain to improve the delay and throughput of the machine-type communication (MTC) systems. There are studies to dynamically adjust the ACB factor by deep learning or prediction algorithms. However, in real systems, the current or next cycle load quantity is not directly known at the base station side. To overcome this difficulty, in this paper, the load is estimated based on Markov chain and then the load of the next cycle is predicted using various neural networks to adjust the ACB factor in advance. Simulation results show that the system performance of this scheme is better than the Markov chain-only and the neural network-only scheme.