Performance Analysis And Optimization On Wireless Carrier Aggregation Systems

To satisfy the increasing requirement of wireless communications, International Telecommunication Union (ITU) has issued a Circular Letter in 2008, to invite submission of candidate technologies for the advanced version of International Mobile Telecommunications (IMT-Advanced) systems, i.e.4G systems. One of the key features of 4G systems delineated by ITU is to support up to 1Gbps peak data rate in downlink. Consequently, the system bandwidth should firstly be extended, even up to 100MHz. As one of the most important techniques for 4G systems, carrier aggregation (CA) can extend the system bandwidth of 4G systems easily and flexibly while guarantee good backwards compatibility, so that reduce its time-to-market. This dissertation focuses on the performance analysis and optimization of the CA based wireless communication systems, also named as wireless CA systems, which is the Orthogonal Frequency Division Multiplexing (OFDM) based downlink system.In wireless CA systems, there are multiple system carriers, which can be aggregated together to form very large system bandwidth. Theses carriers are also called as component carriers (CCs) in wireless CA systems. Intra-band CC aggregation scenario and inter-band CC aggregation scenario are two main scenarios that are considered in the study of this dissertation. In each scenario, the traffic input can be with either single flow or multiple flows. In intra-band CC aggregation scenario, this dissertation firstly analyzes the performance of wireless CA systems under the basic system resource scheduling schemes, then tries to achieve better system performance by designing advanced system resource scheduling schemes. In inter-band CC aggregation scenario, the performance enhancement of wireless CA systems is considered in this dissertation by modifying the system resource scheduling schemes that have been designed in the intra-band CC aggregation scenario.The major contributions of this dissertation include:1. In intra-band CC aggregation scenario with single flow traffic input, two system resource scheduling schemes are designed, termed as Joint Carrier Scheduling (JCS) and Independent Carrier Scheduling (ICS), based on the completely joint and completed independent CC coordinate mechanisms, respectively. Meanwhiel, three user allocation rules are given for ICS, termed as Random Allocation (RA), Alternate Allocation (AA) and Join the Shortest Queue (JSQ). Then the resource utilization of the system is analyzed when JCS and ICS are used using queue theory. The analysis shows that 1) under JCS, the system performance can be better than that of ICS. Furthermore, such performance gain will increase if there are more CCs in the system; 2) the system performance under ICS is relative to the user allocation rule used. Among the three given user allocation rules, the system performance is best when JSQ is used, AA worse and RA worst.2. In intra-band CC aggregation scenario with single flow traffic input, according to the analysis on the system performance under JCS and ICS, a novel dynamic CC coordinate mechanism is proposed and accordingly an advanced system resource scheduling scheme, termed as Dynamic Coordinated Carrier Scheduling (DCCS), is designed, while two methods that are used in DCCS to guide the CCs' collaboration are given. By simulation, it is shown that, if the users can add and remove CCs without delay, using DCCS scheme can make the system achieve the same performance but lower complexity than that of JCS, no matter which kind of user allocation rule and CC coordinate method are used in DCCS. Among the three given user allocation rules, JSQ can make the system complexity lowest.3. In intra-band CC aggregation scenario with multiple flows traffic input, two system resource scheduling schemes are designed, termed as Joint User Scheduling (JUS) and Separated Random User Scheduling (SRUS), based on the completely joint and completed independent CC coordinate mechanisms, respectively. By simulation, it is shown that 1) when the traffic load of the system is larger than 50%, the system performance when SRUS is used is very close to that of JUS; 2) when the traffic load of the system is very small, the system performance under SRUS is far from that of JUS, even up to 100%. According to the simulation results, a so-called Separated Burst Level Scheduling (SBLS) scheme is proposed for the scenario that the system traffic is finite buffer. In SBLS, the connected CC of each user can be changed in burst level, while in SRUS, it is fixed. Meanwhile, SBLS limits the users to receive from only one of the CCs simultaneously, which is the same as that in SRUS. In this way, SBLS can make system achieve better performance than SRUS but at the same time, with acceptable complexity increase.4. In inter-band CC aggregation scenario with multiple flows traffic input, since the CCs that are in the different frequency bands may have distinct radio propagation characteristics, the transmission rates that they can afford the same user are different. Therefore, a new user allocation algorithm is proposed. Simulation results show that when the system traffic is full buffer, comparing with the random user allocation algorithm, the proposed algorithm can make the system perform better. Furthermore, the performance gain of the proposed algorithm to the random user allocation algorithm will increase when the frequency distances among the frequency bands that the CCs locate in become larger. 5. In the scenario that the CCs can be in any frequency band and with any size of bandwidth, and the traffic input is with single flow, the analytical expression of the optimum user allocation weight is derived for ICS using RA, to minimum the average user backlog.