Random access control algorithm guaranteeing diverse QoS requirements of massive machine type communication devices

Abstract

In a massive machine type communication (MTC) environment, a traffic control algorithm is essential to solve a traffic congestion problem. Also, as MTC applications have diverse quality-of-service (QoS) requirements, traffic control should be applied in a way of guaranteeing diverse requirements. In this thesis, first, we consider various delay requirements of machine type communication devices (MTCDs) and propose a priority-based access class barring (PACB) scheme. We divide MTCDs into multiple classes according to their delay requirements and uses a different access class barring (ACB) factor for each class. This thesis describes a numerical model for deriving throughput, drop probability, and access delay. For a realistic environment, we also suggest an estimation algorithm for the PACB scheme, in which a base station (BS) estimates the number of active MTCDs in each class. Simulation results show that the PACB scheme can decrease the delay of a high priority class and improve the throughput and drop probability compared to existing algorithms. Second, we suggest a partially clustered access scheme (PCAS) in which direct-access MTCDs (dMTCDs) and clustered-access MTCDs (cMTCDs) are serviced together. Traffic congestion of a random access link can be resolved by clustering MTCDs and using an aggregator. However, the conventional clustered access scheme (CAS) only considered the environment that only cMTCDs exist. In CAS, as aggregators send access request to the BS when the length of queue exceeds the access threshold, the access delay may increase. To solve such problem, we suggest PCAS and evaluate the performance through numerical analysis. Simulation results show that PCAS has similar traffic control effect to CAS, and the delay requirement of dMTCDs can be satisfied. We also provided an energy consumption model for PCAS, and showed that PCAS can reduce the energy consumption of cMTCDs. Thus, when using PCAS, MTCDs with delay-constrained application should access directly while MTCDs with energy constraint should be clustered. Finally, we suggest the traffic control algorithm which combines ACB and PCAS to guarantee delay and energy consumption requirements simultaneously. As cMTCDs show low energy consumption performance in PCAS, MTCDs with low energy consumption requirements should be clustered. However, some MTCDs may have both low energy consumption and low delay requirement. Thus, we used ACB together with PCAS for better traffic control and to satisfy delay requirements more precisely. We divide MTCDs into four classes considering their delay and energy consumption requirements, and use multiple ACB factors. To obtain the optimal ACB factors to guarantee requirements of high priority classes, deep Q learning algorithm is used. Through simulation, it is shown that by using ACB together with PCAS, the drop probability can be decreased because of the additional traffic control effect of ACB. Also, the probability of satisfying delay and energy consumption requirement can be improved in case of high priority classes.