| 1. IntroductionWith the rapid development of Internet technology, which based on the TCP/IP protocol network congestion has become a bottleneck restricting the development of the network. Congestion will easily made service (QoS) performance declined such as transmission delayed and throughput quality, which severely affected utilization of network resources including bandwidth, cache and so on. Therefore, the network congestion control has always been a hot issue.At present, there are several methods which people do network technology research: analysis, simulation and experimental technique. As the simulation software is relatively simple to operate and easy to implement, so nowadays people toward network congestion control algorithms and the study of the new agreement mostly through simulation software to authentication such as the OPNET, NS-2 network. However, these simulation results have a very strong restrictions of their own and a large number of assumptions, and there were significant differences between the actual network environment. NS2 and Openet now is the best platform for simulation experiments for the international community of the new algorithm and agreement. However, there are essentially different between NS2 and Openet with the actual network environment, for example in many details it is too idealistic to achieve, the simulation results with the actual network environment far from the experimental results. Direct result of these issues is a great deal of research out of the existing network congestion control algorithm of the original intention and objectives and the results obtained is difficult to further in-depth practice. And the research of network control have the same problems with network congestion control algorithms and new agreement study.Therefore, design a comply with the TCP / IP Internet communications protocol, the network parameters controllable in a certain range, and can accurately measure the performance of the network experiments network control platform for network congestion control algorithm, the new agreement, as well as network control study is very necessary. Through the research platform to convergence the center and direction, both to the question of the findings and verify the conclusion, are provide a explore the basis which have important practical significant2. The main research contentDesign network control experimental platform is to let platform used in control and network-based control algorithm and other researches. The goal of this paper is design the network control platform for network congestion control algorithm. For this reason, the work is as followed:(1) Design of the platform. First of all, analysis demand of the platform. The platform is used for network congestion control algorithms, network protocols, as well as web-based control study. Due to various reasons, this paper design network control experiment platform is to give a platform for network congestion control algorithm. Then, determine the design target of platform. According to this article, the certain target is design and implement a experimental platform for network control which compliance with the TCP/IP Internet communications protocol, the network parameters are controllable in a certain range, and can accurately measure the network performance parameters. Finally, develop the platform design. According to the target of the design platform, develop the platform design, the platform is composed of the sending module, the router module and the receiver module three parts (sending modules and receiver modules belonging to different sub-network segment of the LAN). Sending module is responsible for the packet generation and sending; data transmitted arrival of the receiving module by the router module; receiver module is responsible for receiving data. Set the bottleneck link between two router, creating congestion for the congestion control algorithm research.(2) Platform realization. According to the target of the design platform and design scheme, we select the appropriate hardware and software (including the selection of Ipcop router algorithm as a new goal to add), configure the network parameters, set the corresponding parameters of the software to complete the construction of platform.(3) Data processing software development. An important task of Platform is the accurate measurement of network performance indicators. Herein we use Ethereal software for data tracking and the performance measurement work. But if we want to get the packet loss rate, throughput, delay, delay jitter, transient batch queue, the average queue length, the instantaneous queue length and other performance indicators, it needs the analysis experimental data of the platform. Ethereal Network Analyzer has powerful features, but it can not completely give the above performance, this paper has developed a series of data-processing software for these analysis of network performance indicators.(4) Detection platform. Using NS-2 and network control platform, we load RED and DropTail algorithm for the testing of packet loss rate, throughput, delay, queue length, and analyze two types of algorithm performance environment. Then, we do experiments similar to that in the structure of the network control platform. We get the same conclusion basically. It is proved that the design of the platform, design method, network measurement and data processing, hardware and software platform of choice are correct.(5) Research new method of router congestion control algorithm embedded. First of all, we instruct how Linux, Iproute achieved congestion control, as well as how Linux simply communicated with Iproute, and how Ipcop router achieve congestion control. Then, taking congestion control algorithm ARED for example we explain the method of embeding algorithm in the router Ipcop in detail and successfully embedded ARED in soft routing algorithm.(6) Verify a method of add the new algorithm. First of all, using NS-2 simulation software to load RED and ARED algorithm tested packet loss rate, throughput, delay, queue length and other performance indicators, we get an conclusion that ARED reduced queue average and instantaneous queue length and delay time increased throughput and the sudden flow of absorptive capacity compared with the RED. Through a similar experiment in the platform, the conclusion and NS-2 simulation results are basically consistent, which prove that the new algorithm has been added to the router and called, and that the method of adding a router congestion control algorithm is right.3. Next workAs the network control platform is complex for the design and realization, the research of platform is still in the initial stage. Many aspects need to be improved and developed:(1) Optimization experimental methods, the experimental methods and data-processing methods of platform in the current need to spend a lot of manpower and in order to reduce consumption in this area, it need to optimize the experimental means.(2) Add appropriate equipment to the platform used for wireless communications for research. (3) Platform has a large number of data analysis software which provides a lot of convenience for security research, so the platform can be used for the research of offensive and defensive network.(4) Platform has a good scalability, the object, sensors and other control equipment can be added to the platform and then do the Network Control research based on web by the platform.(5) Currently network control platform is mainly used for studying routing congestion control algorithm. However, due to the routing module used by the experimental platform is Ipcop soft routing, sending module and receiver module using the Linux operating system, TCP sending program is compiled based on Linux. And Ipcop and Linux is open source, which can modify the Linux source code and source Ipcop to study congestion control algorithms and the new agreement.(6) At current, the structure of the network control platform is the same laboratory experiments with two different LAN segment of the relatively simple network environment. In order to get closer to the actual and accurate grasp of algorithm performance, it need to verify the experimental algorithm in more complex network environment. The method is to increase the number of modules and expand the experimental scope, set up a number of experimental points to study in the experiment by the Internet or campus network across the country, the realization of the experimental environment will be more complex, more accurately grasp algorithm performance. |
PDF Full Text Request