Research Of Packet Scheduling Algorithm In HSUPA

The essence of scheduling is the competition and allocation for resources, the main purpose of the wireless packet scheduling is to gurantee the fairness between users and satisfies users'demands for quality of service. High Speed Uplink Packet Access (HSUPA) is a new technical specification of Release 6 which made by the third Generation Partnership Project(3GPP) for WCDMA system. In the HSUPA, Node B controlled scheduling which make the scheduling faster and more flexible increases the cell's capacity and reduce delay, also it make the user equipment (UE) have more uplink throughput and less transfer delay. A proper scheduling algorithm for different traffic modle is very important to improve the system's performance.There are three traditional algorithms in HSUPA, they are Max C/I, RR and PF, the Max C/I wants to maximize the system's throughput and RR wants to make sure all the users are absolute fair, so we usually use the two algorithms as evaluation criterias. The PF is the trade-off of them, to make user's proportional fairness. The number of the users becomes more and more, and the business types become more complex so there are some new algorithms to fulfill users'QoS demands, they are M-lwdf, EXP, and Min-GBR algorithms. These algorithm are mainly to satisfy the users'delay demand and gurantee bit rate(GBR)demand and make some improvement based the PF algorithm.We mainly study the performance of the existing algorithms in the HSUPA in our paper, and we also make some simulation and comparisons between them to find the best one which proper for the HSUPA and make some modification for the HSUPA's new specialty. This paper frist study the capacity of the Max C/I, RR and PF for the non-real time traffic model, and find the most proper algorithm in the HSUPA for the non-real time traffic model, also study the performance of the M-lwdf, EXP, Min-GBR for the Video traffic model, then proposed a new scheduling algorithm M-PF for our specific HSUPA system to fulfill our special requirement. In the last, we improved the existing scheduling mechanism for the mixture traffic model in our sytem.