Performance analysis of persistent allocation in cellular systems

Abstract

Reduction of control overhead required for allocating data channel resources has been one of the key metrics for evaluating and selecting radio access technologies. To improve data channel utilization, cellular systems support data channel resource multiplexing with fine granularity for various lengths of packets. However, the fine granularity of resource allocation in a data channel is obtained at the expense of increased control channel overhead. The excessive control overhead for allocation resources to periodically transmitted packets with small payload motivated the development of persistent allocation. Focused on the small length and the periodic generation of voice over internet protocol (VoIP) packet, persistent allocation was originally adopted in cellular networks to reduce the excessive control overhead. Essentially, persistent allocation is applicable not only to VoIP traffic but also to any other traffic types of which packets are generated periodically with small payload. Traditionally, machine-type traffic is featured with aperiodic and sporadic packet generation. Recently, motion control of a machine is identified as one of 5G application scenarios, and traffic characterization of it revealed that both downlink or uplink packets of motion control are periodically transmitted and have small payload. Thus, persistent scheduling is considered to be necessary for resource allocation of motion control traffic. The performance of persistent allocation has been extensively studied in the literature. Since VoIP traffic was the original target of persistent allocation, most of previous studies evaluated the increase in VoIP capacity due to control overhead reduction. Although there were some analytical results, it is worth noting that these studies presented numerical results obtained by system-level simulations. Despite these previous studies, the control overhead reduction by persistent allocation has not been mathematically analyzed in the literature so far. In this dissertation, the performance of persistent allocation for not only VoIP traffic but also motion control traffic is mathematically analyzed. Unlike previous analytical models, a multi-dimensional Markov chain is employed. For VoIP traffic, a continuous-time Markov model is constructed and analytical results based on continuous-time analysis are presented. For motion control traffic, a discrete-time Markov model is constructed and analytical results based on continuous-time analysis are presented, where the discrete-time Markov model is more suitable for performance analysis of cellular systems operating on a frame basis. Performance of persistent allocation is analyzed in terms of blocking probability and utilization of data channel resources. Finally, based on the obtained blocking probability, the maximum achievable control overhead reduction by persistent allocation is analyzed in terms of control overhead reduction ratio. To verity the analysis results, simulation results are compared with analysis results and good agreement is observed by the comparison.