Research Of Internet Congestion Control Mechanism And Stability Analysis | | Posted on:2013-02-22 | Degree:Master | Type:Thesis | | Country:China | Candidate:Q L Chen | Full Text:PDF | | GTID:2218330371461646 | Subject:Computer system architecture | | Abstract/Summary: | PDF Full Text Request | | In recent years, the research of interent congestion control has been rapidly developed. Butthose algorithms which based on heuristic methods are short of valid control theory andmathematical proofs. In additon, these algorithms are very sensitive to the parametersconfiguration. It is easy to show that different parameters configuration can make differentresults. Therefore, the poor scalability of these algorithms can not meet further requirements. Thelatest network congestion control are based on a series of rigorous mathematical proof includingoptimization theory, nonlinear dynamic analysis and so on to design new congestion controlalgorithm adaping for the network.Our research of the stability of congestion control algorithm is based on the pricemechanism and the theory of optimal flow control. Its main work includes:1. The transfer function of REM algorithm is deduced. By using Routh stability criterion, atheoretical analysis of the stability of TCP/REM model and a theoretical guidance in determiningthe parameters of REM algorithm are given.2. A self-tuning REM (STREM) algorithm which changes the control parameterautomatically is proposed. Simulation results show that this mechanism can maintain high linkutilization in a wide variety of network conditions.3. The stability of the primal-dual algorithm for Internet congestion control is investigated inthe presence of heterogeneous round-trip communication delays. Based on the geometricanalysis of frequency response functions of the congestion control system, some sufficientconditions for the local asymptotic stability of the primal-dual algorithm are established. Theseresults explain the main effect of delay in the communication network and have further guidingsignificance on the actual network parameter settings. | | Keywords/Search Tags: | congestion control, active queue managment, optimization theory, networ kdelay, stability | PDF Full Text Request | Related items |
| |
|