| Co-Investigator(Kenkyū-buntansha) |
BABA Ken-ichi Osaka University, Cybermedia Center, Associate Professor, サイバーメディアセンター, 助教授 (60252722)
KOBAYASHI Aki National Institute of Multimedia Education, Associate Professor, 助教授 (30323801)
SAKAI Yoshinori Tokyo Institute of Technology, Global Scientific Information and Computing Center, Professor, 大学院理工学研究科, 教授 (70196054)
IIDA Katsuyoshi Tokyo Institute of Technology, Global Scientific Information and Computing Center, Associate Professor, 学術国際情報センター, 講師 (00332768)
|
|---|
| Budget Amount *help |
¥14,500,000 (Direct Cost: ¥14,500,000)
Fiscal Year 2006: ¥5,700,000 (Direct Cost: ¥5,700,000)
Fiscal Year 2005: ¥5,000,000 (Direct Cost: ¥5,000,000)
Fiscal Year 2004: ¥3,800,000 (Direct Cost: ¥3,800,000)
|
|---|
| Research Abstract |
We proposed a new quality of service (QoS) control algorithm using Diffserv architecture, namely dynamic class assignment (DCA) method. In DCA, routers classify best effort applications, such as WWW and FTP, into several classes based on the difference in their traffic characteristics and assign certain amount of bandwidth for each class ; Multimedia streaming traffic is not statically assigned to a specific class, but routers dynamically choose a class to assign on each new flow arrival. By making routers choose a class which is expected to be the best on each flow arrival with taking into account each class's amount of traffic and characteristics, and the influence of the new flow to the existing traffic in the class, DCA enable to reduce negative effect of streaming traffic to best effort applications and provide the streaming traffic with their desired bandwidth. We also proposed a new admission control method, tentative accommodating and congestion confirming strategy (TACCS), Whe
… More
reas conventional admission control schemes are typically based on the same concept as circuit switching networks, i.e., admission control is done by checking if available resource is enough for a new flow, in TACCS, routers tentatively accommodate incoming flows without checking available resource and check whether or not they cause congestion after the tentative accommodation. By this way, TACCS enables admission control without resource information management or signaling, and it maintains statistical multiplex effect, which is one of the important benefits of packet switching networks. Through simulations and mathematical analysis, we demonstrated that our methods are effective and that we can achieve more scalable QoS control by combining DCA and TACCS. Moreover, we implemented DCA routers and did experiments using them. Through the experiments, we showed DCA achieves to maintain the quality of both stream and best effort applications traffic even where conventional routers degrade both types of traffic. Based on the results, we concluded we accomplished the purpose of this research, i.e., achieving a new QoS control which reduces negative interactions between stream and best effort applications traffic and provide stream traffic with their desired bandwidth. Less
|
