Research On Call Admission Control Mechanism With Self-Similar Traffic Input

With the rapid development of the Internet and the quick increment of the network traffic, more and more new services required strict quality of service (QoS) guarantee have been emerged on the current network,as a result of that IP network must be offered corresponding mechanism for guarantee QoS while improving utilization high-efficiently. Connection admission control (CAC) is one of the methods for QoS guarantee.On the other hand, network traffic processes exhibit ubiquitous properties of self-similarity and long-range dependence(LRD).At present, there are two kinds of mechanisms for QoS-Based CAC, namely model-based CAC and measurement-based CAC. In the model-based CAC research, traditional CAC is nearly nature, while the study on CAC based on self-similar model has just begun, and it will be the research focus in the future.The quality of CAC algorithm determines the comprehensive performance of the network directly.Permitting too much traffic into the network will lead to the fact that some traffic can not meet their QoS requirement;On the contrary, refusing the traffic wrongly will cause the waste of resources and result in low utilization. It has great relationships between CAC and traffic model, and through the study of CAC based on self-similar traffic model and consideration of the traffic characteristic of self-similarity, the efficiency of traffic management mechanism can be improved and the QoS guarantee of self-similar traffic best offered.Self-similar traffic model-based and network calculation-based CAC schemes are deeply studied. The CAC algorithms proposed in this thesis can best meet uers' QoS requirement and achieve higher utilization through fully consideration of the characteristic of self-similarity of network traffic. The main research works and achievement of the thesis are as follows:1) The QoS architectures are analyzed and the methods of QoS parameters calculation based on self-similar traffic are studied.2) Effective bandwidth is the main basis of CAC.The way to compute effective bandwidth based on FBM andα-stable self-similar process is researched, and the variances of effective bandwidth with other characteristic exponent are studied through theoretical analysis and simulation.CAC based on effective bandwidth is proposed, and the feasibility and effectiveness of CAC are proved through simulation.3) Traffic shaping is the most common use of traffic conditioning technology. Parameters computing methods of token bucket used to shape the self-similar traffic based on FBM and LFSN are studied in detail.4) Self-similar traffic CAC based on network calculus is deeply studied. Based on network calculus, the implementation method for resource reservation recommended by IETF and its problem are studied in detail. For solving this problem, two optimal allocation schemes for network resource constrained on the arrival curve, maximum delay and buffer size are proposed, and the desired rate and buffer size of these schemes are derived.