Dynamic bandwidth scheduling and burst construction algorithm for downlink in (4G) mobile WiMAX networks

Abstract

Advanced wireless systems, also called fourth generation (4G) wireless systems, such as Mobile Worldwide interoperability for Microwave Access (WiMAX), are developed to provide broadband wireless access in true sense. Therefore, it becomes mandatory for such kind of systems to provide Quality of Service (QoS) support for wide range of applications. In such types of systems, wireless base stations are responsible for distributing proper amount of bandwidth among different mobile users, thus satisfying a user’s QoS requirements. The task of distributing proper amount of bandwidth rests upon a scheduling algorithm, typically executed at the base station. 2G and 3G wireless systems are able to provide only voice, low data rate, and delay insensitive services, such as Web browsing. This is due to the lack of development in digital modulation and multiple access schemes, which are two major aspects of physical layer of these systems. Digital modulation is used to combat with location-dependent channel errors which get introduced in the data transmitted by base station on a wireless channel to a mobile station. Hence, different locations of every mobile station in a cell coverage area require different modulation and coding schemes for error-free transmission. Link adaptation is a technique that makes the use of variable modulation and coding schemes possible, according to varying location of mobile stations. This technique is used by 4G systems to achieve error free transmissions. 2G and 3G systems are not capable of achieving error-free transmissions in many cases due to significantly fewer or no choice of modulation and coding schemes for different locations of mobile stations. In such cases, most of the time, wireless channel is either error-prone or error-free for mobile station. Scheduling algorithms developed for 2G and 3G systems focussed on providing long term average rate requirements of users, which are satisfied at the expense of zero transmission for mobile users experiencing bad or error prone channel. This approach was adopted to achieve efficient use of wireless channel capacity. This was the best approach adopted by majority of scheduling algorithms because delay sensitive applications were not supported in such systems and hence bounded delay was not a matter of concern. Hence, the majority of the algorithms focussed on providing long term average rate requirements while maximizing cell throughput. This helped in making efficient use of wireless channel capacity at the expense of zero transmission for mobile users experiencing bad channel and compromising delay performance. These approaches, however, will not be suitable for 4G systems as such systems support wide range of applications ranging from delay-insensitive to highly delay-sensitive. Hence in this thesis, a dynamic bandwidth scheduling algorithm called Leaky Bucket Token Bank (LBTB) is proposed. This algorithm exploits some advanced features of 4G systems, like link adaptation and multiple access scheme, to achieve long term average rate requirements for delay-insensitive applications and bounded delay for delay-sensitive applications. Advanced features of 4G systems also bring more challenges. One such challenge is Orthogonal Frequency Division Multiple Access (OFDMA), a multiple access scheme deployed in 4G systems. In OFDMA, scheduled data for different mobile stations is packed into bursts and mapped to a two dimensional structure of time and frequency, called OFDMA frame. It has been observed that the way bursts are mapped to OFDMA frame affects the wakeup time of mobile stations receiving data and therefore causes power consumption. Wakeup time is the time duration in OFDMA frame for which the mobile station becomes active. Since OFDMA frame is a limited and precious radio resource, the efficient use of such radio resource is necessary. Efficient use requires that the wastage of such radio resource be minimized. Hence in this thesis, a burst construction algorithm called Burst Construction for Fairness in Power (BCFP) is also proposed. The algorithm attempts to achieve fairness in power consumption of different mobile stations by affecting their wakeup time. It also attempts to minimize wastage of radio resources. For comparing the performance of joint proposed algorithms (LBTB+BCFP), the proposed burst construction algorithm (BCFP) is joined to the two other existing scheduling algorithms namely: Token Bank Fair Queuing (TBFQ) and Adaptive Token Bank Fair Queuing (ATBFQ). TBFQ is an algorithm developed for 3G wireless networks whereas, ATBFQ is an extension to the TBFQ and is developed for 4G wireless networks. Therefore, the performance of the proposed algorithms jointly together (LBTB+BCFP) is compared with the joint TBFQ and proposed burst construction algorithm (TBFQ+BCFP), as well as joint ATBFQ and proposed burst construction algorithm (ATBFQ+BCFP). We compare the performance in terms of average queuing delay, average cell throughput, packet loss, fairness among different mobile users, fairness in average wakeup times (average power consumption), and fraction of radio resources wasted. The performance of proposed burst construction algorithm (BCFP) is also compared with Round Robin algorithm in terms of fairness in average power consumption as well as fraction of radio resources wasted, for varying number of users.