Dynamic Spectrum Allocation for mMTC Devices Using SCMA-Q-Learning with ACB Overload Control

With the development of 5G technology and wireless networks, the limited spectrum resources and increasing number of Machine-Type Communications (MTC) devices in massive Machine-Type Communications (mMTC) networks have led to the problem of Random Access Channel (RACH) congestion. Sparse Code Multiple Access (SCMA) is a widely used method for random access (RA), which can effectively improve spectrum utilization efficiency. However, using the SCMA method for RA can result in the codebook collision. This paper presents a SCMA-Q-learning method to deal with Low Earth Orbit (LEO) satellite communication networks congestion. This method allow MTC devices to dynamically select the optimal codebook and time-slot-group based on the constantly changing environment. Access Class Barring (ACB) technology is employed in overload control. When the system is overloaded, the ACB factor adjusts its size based on the number of accessed devices. Simulation results show that compared with other existing methods, our method can effectively improve the throughput of the entire system.