Research On Qos Measurement And Optimization In Wimax Wireless Communication Networks

The WiMAX (Worldwide Interoperability for Microware Access System) industry has seen rapid growth in recent years, and many experimental networks have been established around the world. The IEEE 802.16 standard suite has defined five types of service flows, and each connection has the corresponding user flows. The standard has defined detailed QoS parameters for different user flows, but it does not specify the algorithms for admission control, scheduling, and congestion control. A distinguishing feature of WiMAX system is its perfect QoS guarantee, and the algorithms for scheduling, admission control, and congestion control have significant impact on QoS guarantees. Based on measurement of a variety of parameters and their dynamic trends, this dissertation studies how to improve user QoS in wireless communication networks. The main contributions of this dissertation are as follows:(1) The research on low header-overhead scheduling for real-time media flows. In the scheduling for real-time media flows, subheader overheads grow with the number of active SS (subscribe stations), thus introducing significant amount of header overhead. To address this issue, this dissertation proposes an effective, low-overhead scheduling algorithm to reduce uplink header-overhead. Low-overhead scheduling algorithm first schedules SSs with the earlier-deadline-first (EDF) algorithm. For the remaining bandwidth, this dissertation uses a Batch Markovian Arrival Process (BMAP) to predict the arrive rate of real time media flows, and it uses Newton's interpolation polynomial function to predict the response time from BS; based on the difference between the predicted and actual values, Low-overhead scheduling algorithm adaptively adjusts errors, and it precisely predicts the bandwidth that should be allocated to uers and pre-allocate the bandwidth for users. Low-overhead scheduling algorithm reduces the number of users in a frame, thus reducing the overhead in WiMAX uplink scheduling. Simulation results have demonstated that, compared with the existing algorithms, a Low-overhead scheduling algorithm not only ensures QoS but also significantly reduces IP header overheads.(2) The research on uplink scheduling problems. By analyzing existing scheduling algorithms and based on the QoS requirements of different user flows, this dissertation proposes a hierarchical scheduling algorithm. For uplink scheduling of real-time media flows, this dissertation proposes the urgent-DRR (UDRR) scheduling algorithm; the UDRR scheduling algorithm modifies and extends the DRR scheduling algorithm by considering mechanisms that have been employed by the network and the video applications and by adding weight to delay urgencies of service flows. Compared with the traditional RR and WRR algorithms, UDRR scheduling algorithm can reduce end-to-end delay and delay jitter and improve system through for bursty video services, thus further meeting the requirements of real-time transmissions.(3) The research on user admission control and congestion control in WiMAX networks. Based on real-time measurement of user flows and the dynamically estimated QoS parameters, this dissertation proposes a dynamic admission control algorithm based measured QoS parameters. In the mean time, this dissertation proposes a congestion control algorithm based on the impact of wireless link errors on upper-layer performance. This algorithm reduces congestion by determining the degree of network congestion through a (M, N) statistical measurement algorithm based end-to-end data loss classification. The simulation results have shown that the proposed admission and congestion control algorithms can effectively improve user QoS and system performance.(4) The research on scheduling problems. The WFQ scheduling algorithm has limitation in the presence of bursty real-time media data. To address the limitations, this dissertation proposes a measurement-based WFQ scheduling algorithm. Evaluation results have shown that the measurement-based WFQ algorithm is robust to bursty media data and outperforms the WFQ algorithm for the download of bursty real-time media.