Research And Simulation Of Queue Management Technologies Of IP QoS

With the evolution of Internet and IP network technology, networks based on IP technologies have not been the ones that just have a few simplex traffics running any more. Many of them have carried many different kinds of traffics, with different demands of quality of service, which consist of kinds of data payload, such as text, voice, video and images etc. Traditional IP technologies cannot meet the complex and diverse requirements of quality of service. In this context, networks must rely on IP Qo S technologies to guarantee the quality of service for different traffics. Queue management is the important tool for implementing policy of resource allocation and Per-Hop Behavior in Diff-Serv architecture. With the increasing complexity of network application scenarios, however, queue management scheme with simplex structure cannot fulfill the combination of different demands of quality of service. Therefore, it's quite necessary to study queue management technologies that is enough flexible to accommodate itself with that complex requirements.The paper firstly tries to explain a kind of important Qo S architecture, i.e. the Differentiated Service Architecture, and describe the summary of its architecture and its core components, then investigates the critical technologies used to implement these components, mainly focusing on traffic policing, traffic shaping, and packet queuing, scheduling, congestion avoidance and so on. After summarizing the characteristics of these critical techniques and taking typical scenarios of network applications into account, the paper designs a compound-structured queue management scheme from the point of view of service quality measures. This scheme provide preemptive scheduling privilege to traffics demanding low latency and share the remaining schedule chances between active traffics that demands specific bandwidth in a weighted fair manner according to max-min fair share principle. The scheme tries to fulfill the complex requirements scenarios, e.g. the situation where traffics of low-latency and the ones demanding specific bandwidth co-exist.Finally based on the OPNET Modeler platform, the paper implements the scheme and run simulation tests of it in contrast with that of the built-in queue management scheme of Modeler. The analysis and comparison of simulation results show that the compound-structured queue management scheme is more accommodated with complex requirements of quality of service.