Research Of Combined Methodology For Measuring Of Available Bandwidth And Locating Internet Bottleneck

Presently, most of the network measurement tools are based on activemeasurement method. They bring heavy traffic load which affects the normaloperation of the network and hurts measurement accuracy. With in-deep studying ofthe existing methods and tools for measuring of available bandwidth and locatingInternet bottleneck,a new measurement scheme iPathneck based on Pathneck isproposed to perform better than Pathneck on measuring available bandwidth by usingprobe rate model and to improve the locating bottlenecks. iPathneck achieves abalance between the measurement accuracy and the probing overhead by combiningtasks to measure available bandwidth and to locate Internet bottlenecks. The premiercontent of this paper is as follows.Study and compare the existing Internet tools and methods for availablebandwidth measuring and bottleneck locating. Pathneck performs better than otherson bottleneck locating. It can accurately locate the Internet bottleneck with littlepackets probing. But this tool has some shortcomings. On one hand, it can only roughestimates the upper and lower bounds of available bandwidth with its imperfectmeasurement mechanism and is sensitive to the packets’ space and to the burstbackground flow. On the other hand, the tool reduces the accuracy of bottlenecklocating by ignoring the impact of measurement packets on the length change ofrecursive packets train. Then iPathneck improves to resolve the problems of Pathneck.First of all, iPathneck fallows the probe rate model to measure the availablebandwidth to solve the measurement mechanism problem fundamentally. Probe ratemodel isn’t sensitive to the packets’ space and the burst background flow. So,iPathneck improves the accuracy and the robustness of the available bandwidthmeasurement method.Secondly, probe rate model will increase the probe packets injection. To solvethis problem, iPathneck send assistance packets train which is of less overhead thanrecursive packets train assisting the last to search available bandwidth. This methodneither affects the accuracy of bottleneck locating nor significantly increases theprobe packets injection. And the algorithm converges quickly because of using ternaryalgorithm to determine the send rate. And iPathneck increase the calibrationmeasurement part to correct result of the available bandwidth measuring and internet bottleneck locating.Finally,to avoid the effect of measuring packets on the length of recursivepackets train, iPathneck will determine the candidate choke node according to thechanges of gaps between the measurement packets instead of the gaps between themeasurement packets as Pathneck used. This will reduce the measuring errors.Experiments has be done to verify the accuracy and performance of availablebandwidth measuring and of Internet bottleneck locating. Experiments showed theaccuracies of both measurements achieve the requirement of reasonableness.iPathneck measures in a short time and with a small amount of probe traffic. Testverifies that iPathneck achieved a balance between measurement accuracy and probeoverhead by merged measurement tasks to reduce the probe packets injection intonetwork.