Research On Scheduling And Load Balancing Based On Service Characteristic Identification

Currently, with the development of wireless communication technology and the increment of services, people hope whenever and wherever wireless networks can be accessed in order to obtain excellent service experience. Although, the development and use of novel technology has expanded the dimension of wireless resource and improved the resource usage, the wireless resource is rare, for it can't satisfy all users'requirement of subjective experience. Therefore, in this paper, we research traffic characteristic identification method suitable for wireless network, and further research scheduling and mobility load balancing based on services'properties, so as to guarantee services'QoS of wireless communication networks.Firstly, we study traffic characteristic identification algorithm, and propose real-time traffic classification method and conversational video service's frame information identification method suitable for wireless network. The proposed real-traffic classification method uses few packets'statistical characteristic and selects a group of statistical parameters which are insensitive to network situation as the feature vector of k-NN algorithm in order to identify traffic type. Therefore, the classification method can obtain ideal identification results and robustness. We further study the feasibility that the real-time traffic classification method was applied to IEEE 802.16 system, and evaluate the classification results'influence on network performance. In addition, by analyzing the coding properties and network transmission properties of conversational video service, we propose the GOP pattern identification method to obtain video frames'information instantaneously and efficiently, which can be used for base station to optimize the wireless conversational video traffic.Secondly, aiming to conversational video services in 3GPP LTE system, we propose an adaptive scheduling algorithm based on video frame identification. The scheduler, which takes into account frame weightiness, user's channel quality and system load for conversational video services, can adaptively select video frame based on the relationship between frame tolerable delay and time window, so as to obtain higher performance of average video quality and system throughput by preferentially transmitting high priority frames and avoiding unnecessary losses of low priority frames. Finally, we propose a QoS guaranteed mobility load balancing scheme, which is based on the MLB's influence on different services, in order to improve users'experience in hot or congested cell. The proposed scheme can make the handover be different by taking into account services'characteristic and mapping the services into corresponding service hysteresis parameters to preferentially handover the users with non real-time services and avoid handover the users with real-time during mobility load balancing, so as to degrade the influence on the QoS experience of users caused by load balancing. The simulation results show that the proposed MLB scheme can improve the percentage of satisfied users with real-time traffic, and reduce handover times caused by MLB.