| No | 236234 |
|---|---|
| 標題（和） | A Modified Excursion Detection Algorithm for Efficient Measurement of Available Bandwidth Along End-to-End Network Path |
| 標題（英） | A Modified Excursion Detection Algorithm for Efficient Measurement of Available Bandwidth Along End-to-End Network Path |
| 研究会名（和） | 通信方式 |
| 研究会名（英） | CommunicationSystems |
| 開催年月日 | 2015-07-02 |
| 終了年月日 | 2015-07-03 |
| 会議種別コード | 5 |
| 共催団体名（和） | |
| 資料番号 | CS2015-14 |
| 抄録（和） | In this paper, we have developed a modified excursion\r\ndetection algorithm from the one way queuing delay faced\r\nby each packet in a train of packets along the end to end network\r\npath. Pathchirp\'s original excursion detection algorithm has been\r\nmodified and our previous work NEXT has been extended for\r\nbetter estimation of the available bandwidth (AB). The main\r\nprinciple is that, if a certain packet?s rate is below the AB along\r\nthe network path, then no queuing delay occurs and vice versa.\r\nBut due to the bursty arrival of cross traffics, a sudden increase\r\nof queuing delays of packets (called excursion) occur for a short\r\nperiod of time in the router even though the packet?s rate is\r\nmuch below the AB of the tight link. So, to detect these sudden\r\nincrease of queuing delays and to filter out them is the main\r\npurpose of the excursion detection algorithm. We have classified\r\nwhether a certain packet lies inside the excursion region or not\r\nbased on the queuing delay of it as well as all other previous\r\npacket?s queuing delay. If a certain packet?s queuing delay is less\r\nthan the average of the queuing delay faced by all other packets\r\nbefore it, then this packet?s queuing delay is not considered inside\r\nthe excursion region as well as the corresponding rate is not\r\ntreated as the turning point (available bandwidth) of the queuing\r\ndelay signature. Simulation results reveal that, our extended\r\nidea achieves better performance than Pathchirp and overcomes\r\nunderestimation problem in many cases. |
| 抄録（英） | In this paper, we have developed a modified excursion\r\ndetection algorithm from the one way queuing delay faced\r\nby each packet in a train of packets along the end to end network\r\npath. Pathchirp\'s original excursion detection algorithm has been\r\nmodified and our previous work NEXT has been extended for\r\nbetter estimation of the available bandwidth (AB). The main\r\nprinciple is that, if a certain packet?s rate is below the AB along\r\nthe network path, then no queuing delay occurs and vice versa.\r\nBut due to the bursty arrival of cross traffics, a sudden increase\r\nof queuing delays of packets (called excursion) occur for a short\r\nperiod of time in the router even though the packet?s rate is\r\nmuch below the AB of the tight link. So, to detect these sudden\r\nincrease of queuing delays and to filter out them is the main\r\npurpose of the excursion detection algorithm. We have classified\r\nwhether a certain packet lies inside the excursion region or not\r\nbased on the queuing delay of it as well as all other previous\r\npacket?s queuing delay. If a certain packet?s queuing delay is less\r\nthan the average of the queuing delay faced by all other packets\r\nbefore it, then this packet?s queuing delay is not considered inside\r\nthe excursion region as well as the corresponding rate is not\r\ntreated as the turning point (available bandwidth) of the queuing\r\ndelay signature. Simulation results reveal that, our extended\r\nidea achieves better performance than Pathchirp and overcomes\r\nunderestimation problem in many cases. |
| 収録資料名（和） | 電子情報通信学会技術研究報告 |
| 収録資料の巻号 | Vol.115, No.123 |
| ページ開始 | 27 |
| ページ終了 | 32 |
| キーワード（和） | Available Bandwidth,Excursion Detection,Probe Rate Model,Queuing Delay Signature,NEXT |
| キーワード（英） | Available Bandwidth,Excursion Detection,Probe Rate Model,Queuing Delay Signature,NEXT |
| 本文の言語 | ENG |
| 著者（和） | Anup Kumar Paul |
| 著者（ヨミ） | |
| 著者（英） | Anup Kumar Paul |
| 所属機関（和） | KDDI R&D Laboratories Inc. |
| 所属機関（英） | KDDI R&D Laboratories Inc. |
| 著者（和） | Atsuo Tachibana |
| 著者（ヨミ） | |
| 著者（英） | Atsuo Tachibana |
| 所属機関（和） | KDDI R&D Laboratories Inc. |
| 所属機関（英） | KDDI R&D Laboratories Inc. |
| 著者（和） | Teruyuki Hasegawa |
| 著者（ヨミ） | |
| 著者（英） | Teruyuki Hasegawa |
| 所属機関（和） | KDDI R&D Laboratories Inc. |
| 所属機関（英） | KDDI R&D Laboratories Inc. |