| Providing Quality of Service (QoS) differentiation and guarantees in wireless networks is a challenging task due to the special problems of wireless communications, such as link variability and limited bandwidth. New technology needs to be developed to tackle these problems.; Wireless packet scheduling, a key component of QoS provisioning in wireless data networks, is the major focus of this dissertation. Although many wireless packet scheduling algorithms have been proposed in recent years, some issues remain unresolved. We study packet scheduling issues in both TDMA-based and CDMA-based wireless networks and propose three novel wireless scheduling algorithms applicable to different link and network models. First, we design a scheduling algorithm for TDMA networks with wireless links being modeled as having two states. The algorithm is shown to distribute excess bandwidth effectively, strike a balance between effort-fair and outcome-fair, and provide delay bound for error-free flows and transmission effort guarantees for error-prone flows. Second, extending the previous work, we propose a novel wireless scheduling algorithm applicable to wireless links with multiple states. For delay-sensitive flows, the algorithm is capable of providing statistical delay violation bounds. For best-effort flows, we propose a new notion of fairness, called long-term link-quality-weighted outcome-fair. The algorithm balances between bandwidth efficiency requirement and fairness requirement, and guarantees minimal goodput levels for best-effort flows. Third, an adaptive packet scheduling algorithm for cellular CDMA networks is proposed. The algorithm guarantees packet deadline and average data rate under the assumption of perfect power control. It is also power-efficient because it takes into account the varying channel conditions.; Due to the special characteristics of wireless mobile environment, fixed level of service guarantees and fixed level of resource allocation, commonly used in wireline networks, are not suitable and viable in wireless networks. In the last part of our dissertation, we propose a general utility-oriented adaptive QoS model for wireless networks and establish a framework for formulating the bandwidth allocation problem for users with time-varying links. Based on the framework, a high-level adaptive bandwidth allocation scheme, which guarantees utility-based QoS, ensures long-term fairness, and achieves high bandwidth efficiency, is designed. |
