Enhancing Active Queue Management Techniques for Improving Network Performance in Congestion Problem
As a result of the considerable development of Internet technology in numerous applications, Internet applications currently require high-speed router 'buffers to quickly transmit data to their potential recipients. Congestion is a crucial problem associated with the performance of Internet app...
Saved in:
| Summary: | As a result of the considerable development of Internet technology in numerous applications, Internet applications currently require high-speed router 'buffers to quickly transmit data to their potential recipients. Congestion is a crucial problem associated with the performance of Internet applications. Congestion occurs at a particular router buffer when the demand for network resources exceeds the available resources. Consequently, network performance declines. To achieve good network performance, router buffers should be managed with a congestion control method, such as active queue management (AQM). Several AQM methods have been proposed to overcome the congestion problem. However, they failed to show improvement in the network performance. In this thesis, a comprehensive review of the literature on congestion control is conducted. A new method, called the gentle BLUE (GB) method, is also proposed. The GB method, which is based on the BLUE method, detects congestion before the router buffer overflowed, thereby avoiding congestion. The proposed method also reduces the parameter settings by providing a dynamic mechanism for calculating the dropping probability (DP) based on the status of the queue length (ql). The proposed GB method is simulated, and the obtained results are compared with those of existing AQM methods. The properties of packet arrival traffic, such as burstiness and correlation, should also be considered inputs in evaluating AQM methods. Therefore, discrete-time performance analysis is conducted for the GB method using the two-state Markova modulate Bernoulli process (MMBP- 2) to model the queuing process and deal with traffic properties. The resulting method is called GB-MMBP-2. A discrete-time analytical model is adopted to Validate the performance of the proposed GB simulation. This validation is achieved when similar performance measure results are obtained for GB simulation and the GB analytical model. Finally, a congestion control technique based on the GB method is proposed using fuzzy logic (FL) to control congestion at router buffers at an early stage and reduce the dependency of the GB method on its parameters. The GB method using FL (GBFL) adopts ql and delay rate as input linguistic variables for an FL system to produce a single output (DP). Compared with existing methods, the proposed GB method provides improved mean queue length, delay, and packet loss in case of heavy congestion. Moreover, the GB-MMBP-2 method provides the best performance result in terms of mean queue length, delay, packet loss, and DP under bursty and correlation traffic, particularly when heavy congestion occurs. The results of the analytical model are compared with those of the simulator to validate and prove that the obtained simulator results are correct and that the simulator is working properly. The GBFL method offers better mean queue length, delay, and packet loss under light and heavy congestion compared with REDDI and GREDFL. |
|---|