Research On Techniques Of Network Congestion Control For QoS

With the rapid development of the Internet, even if the current network bandwidth has persistently being improved, the new applications continue to expand, making the rapid growth of network traffic as a network congestion, restricting the development of network applications. A direct result of network bandwidth congestion, delay and throughput of other network QoS index down, the lower utilization of network resources, making it difficult to provide good service quality assurance in the area of the network parameters on the particular sensitivity and tolerance demanding business applications. Thus, it has a significance to solve the problem of net congestion and quality of the net service effectively.Network congestion control and quality of service are two closely related research topics, they are also one of the most difficult hotspots in the area of network research. The network congestion is one of the important factors that affects the QoS guarantee. As a crucial method to avoid and control network congestion, it is not only the critical factor of network performance and efficiency improvement, but also the premise and basis to improve network QoS. In this paper, We use the network congestion control as a starting point to study the fair and effective control technology, and support network to provide certain QoS guarantees. On the basis of systematic summarization of up to date network congestion control work, the primary work of the dissertation and innovations are as follows:1.Provide a strategy based on the bandwidth fair implementation of model control. It has been a research focus in the issue of the bandwidth fairness under the differentiated services network in the forwarding process, many elements influence this fairness such as packet size, RTT, target rate and the polymer contained in a single stream flow quantity. To ensure the fairness of the bandwidth issue relys on the network nodes at the boundary marking strategy for the implementation of the packet core network elements and nodes in the implementation of the queue management strategy. In this paper, model-based packet marking rate control strategy is aimed at dealing with the bandwidth between the different streams together to share the fairness. In order to achieve a fair bandwidth allocation, we regulate the rate to change the probability of a different color marker through model-based dynamic feedback in the loan of packet loss rate of the local labeling algorithm. This method's advantage is that it can better predict the source application TCP traffic conditions on the reaction behavior of the network and tag the engine accordingly.Compared with other labeling algorithm, simulation results show that an interactive model-based control method could be more accurate than other methods for each hop packet mark that better predicts the quality of service performance. Meanwhile, the separation mechanism of packet given effectively solves the fairness problem between the flows of response and non-response.2.Provide a congestion control mechanism under heterogeneous network. Analyze the TCP's represented research achievements and current situation problems under heterogeneous environment, according to the shortage of the TCP under heterogeneous network, we design the dual-based aimd algorithm of TCP congestion control algorithm based on the iven bandwidth estimation and its own in the cable vertical handoff in heterogeneous network environment, the results validate the effectiveness of this mechanism under simulation analysis.3.Provide a network service control strategy based on the weak retention mechanism. The growth of network bandwidth and other resources are still unable to adapt to the rapid growth of the business flow, there is still a challenge on how to dynamically control the quality of service priority Service. In this paper, according to the characteristics of the limited network resources, through the proposed network-based control of non-guaranteed flow of traffic to obtain the necessary to ensure the quality of service requirements. The number of non-guaranteed flow of traffic and to ensure that the delay and jitter between the model, the model is controlled through a competitive non-guaranteed flow of traffic in order to obtain the required delay and jitter guarantees, and gives a robust estimation and control of the whole design and analysis. Compared with other well-known scheduling algorithms such as CBQ,WFQ,RIO and etc, we propose a control mechanism of the effectiveness of network services and advantages.4.Proposed a global approach to dynamic allocation of network resources. Bearer network to adapt to changes in the volume of business to take the uncertainty of the dynamic adjustment of network resource allocation strategy is still a difficult research, research in the full reference, based on the unequal distribution of resources for the link caused by network congestion and network quality of service difficult to ensure the problem, a resource allocation strategy based on ant colony optimization algorithm and dynamic. In the analysis of ACO algorithms in the realization of the principle of logical topology reconfiguration of network resources based on the ant colony to find the best path and with different types of ants, the inherent characteristics of mutually exclusive, it is easy to link and network in the shortest link resource trade-off between the load. The results show that the algorithm more frequently on the business case of traffic requests, which reflects the dynamic balance of reliability and service traffic flow arrival rate characteristics of the superiority and effectiveness.