基于多Agent流媒体传输网络拥塞控制机制的研究
|
摘要
Internet的迅猛发展和普及为流媒体业务发展提供了强大的市场动力。流媒体技术广泛用于多媒体新闻发布、网络广告、在线直播、电子商务、远程教育、实时视频会议等互联网信息服务的方方面面。它结束了传统Internet只能显示文字和图片的时代,开创了集声音、视频、文字及图像于一体的新纪元。
流媒体传输技术由于自身传输时延敏感性,半可靠性,传输速率的平稳性等特点,在实际使用中通常采用UDP/IP协议进行传输。但UDP缺乏拥塞控制机制,拥塞发生时它不会降低数据发送速率,这样会引发两种情况:其一, UDP占用的资源将远高于TCP流等响应流所占用的资源,造成了严重不公平性;其二,它将导致拥塞情况的进一步恶化,甚至引发拥塞崩溃,严重影响流媒体服务质量。因此,必须对流媒体传输引入拥塞控制机制。传统拥塞控制算法是通过丢弃数据包后,使发送端通过收到大于等于三个重复的ACK或者重传计时器超时的方式隐式的判断拥塞的发生。本文将利用Agent技术实现拥塞的检测、控制和管理,并提出一种改进的拥塞控制算法Vegas-A+。当拥塞发生时,检测拥塞Agent将实时监测网络并报告拥塞的发生,并由控制拥塞Agent根据网络拥塞情况选择合适的拥塞控制算法来缓解或消除拥塞。利用Agent之间的协作,降低丢包率,提高网络的利用率。
Agent是一个具有自适应性和智能性的软件实体,能代表用户或者其它程序,以主动服务的方式完成一项任务。它具备自主性、反应性、社会性、主动性和适应性等。本文提出用检测Agent来完成报告拥塞发生的功能;控制Agent来完成减轻或消除拥塞的功能;黄页Agent来实现Agent服务公布、服务查询和服务订阅的功能。各个Agent之间相互协作形成一个多Agent系统为网络提供服务。在对Agent功能的分析和设计基础之上,利用JADE开发平台开发多Agent系统使之实现拥塞控制功能。
The rapid development and popularization of Internet provide powerful market power for streaming media business development. Streaming media technology is widely used in all aspects of Internet information services, such as multimedia news release, network advertising, online live, e-commerce, distance education, real-time video conference, etc. It ends the era of traditional Internet which only displayed words and pictures, and marks an epoch in integration of voices, videos, words and pictures.
Because of its transmission delay sensitivity, half reliability, the steadiness of transmission rate, streaming transmission technology adopts UDP/IP protocol for transmission. When congestion occurs, the UDP will not reduce data sending rate for lacking of congestion control mechanism, which will trigger two situations: 1) resources occupied by UDP will be far higher than that occupied by TCP flow, which will cause serious unfairness. 2) It will cause congestion conditions further deterioration, and even cause congestion collapse, which will seriously influence streaming service quality. Therefore, congestion control mechanism must be introduced in streaming transmission. The traditional congestion control algorithms judge the congestion occurrence through the following ways: when packet is discarded, sender will receive greater than or equal to three repeated ACKs, or timeout of retransmission timer. This article will use the Agent technology to realize congestion inspection, congestion control and congestion management, and also puts forward an improved congestion control algorithm Vegas-A+. The congestion inspection Agent will inspect congestion timely and report when congestion occurs. The congestion control Agent will select suitable congestion control algorithm to control congestion according to the reports from congestion inspection Agent. Using the collaboration of all agents to reduce the packet loss rate and increase network utilization.
Agent is an adaptive and intelligent software entity, which can represent user or other software to finish a task in a active service mode. It is autonomy, reactive, sociality, initiative, and adaptable. In this article, inspection Agent will inspect the congestion in the network and send reports, the control Agent will reduce or remove congestion, the yellow page Agent will provide service publish, service search, service subscription. The cooperation of all agents forms a multi-agent system providing services for networks. Based on the analysis and design of all agents’functions, the multi-agent system is developed for solving the network congestion using JADE platform.
流媒体传输技术由于自身传输时延敏感性,半可靠性,传输速率的平稳性等特点,在实际使用中通常采用UDP/IP协议进行传输。但UDP缺乏拥塞控制机制,拥塞发生时它不会降低数据发送速率,这样会引发两种情况:其一, UDP占用的资源将远高于TCP流等响应流所占用的资源,造成了严重不公平性;其二,它将导致拥塞情况的进一步恶化,甚至引发拥塞崩溃,严重影响流媒体服务质量。因此,必须对流媒体传输引入拥塞控制机制。传统拥塞控制算法是通过丢弃数据包后,使发送端通过收到大于等于三个重复的ACK或者重传计时器超时的方式隐式的判断拥塞的发生。本文将利用Agent技术实现拥塞的检测、控制和管理,并提出一种改进的拥塞控制算法Vegas-A+。当拥塞发生时,检测拥塞Agent将实时监测网络并报告拥塞的发生,并由控制拥塞Agent根据网络拥塞情况选择合适的拥塞控制算法来缓解或消除拥塞。利用Agent之间的协作,降低丢包率,提高网络的利用率。
Agent是一个具有自适应性和智能性的软件实体,能代表用户或者其它程序,以主动服务的方式完成一项任务。它具备自主性、反应性、社会性、主动性和适应性等。本文提出用检测Agent来完成报告拥塞发生的功能;控制Agent来完成减轻或消除拥塞的功能;黄页Agent来实现Agent服务公布、服务查询和服务订阅的功能。各个Agent之间相互协作形成一个多Agent系统为网络提供服务。在对Agent功能的分析和设计基础之上,利用JADE开发平台开发多Agent系统使之实现拥塞控制功能。
The rapid development and popularization of Internet provide powerful market power for streaming media business development. Streaming media technology is widely used in all aspects of Internet information services, such as multimedia news release, network advertising, online live, e-commerce, distance education, real-time video conference, etc. It ends the era of traditional Internet which only displayed words and pictures, and marks an epoch in integration of voices, videos, words and pictures.
Because of its transmission delay sensitivity, half reliability, the steadiness of transmission rate, streaming transmission technology adopts UDP/IP protocol for transmission. When congestion occurs, the UDP will not reduce data sending rate for lacking of congestion control mechanism, which will trigger two situations: 1) resources occupied by UDP will be far higher than that occupied by TCP flow, which will cause serious unfairness. 2) It will cause congestion conditions further deterioration, and even cause congestion collapse, which will seriously influence streaming service quality. Therefore, congestion control mechanism must be introduced in streaming transmission. The traditional congestion control algorithms judge the congestion occurrence through the following ways: when packet is discarded, sender will receive greater than or equal to three repeated ACKs, or timeout of retransmission timer. This article will use the Agent technology to realize congestion inspection, congestion control and congestion management, and also puts forward an improved congestion control algorithm Vegas-A+. The congestion inspection Agent will inspect congestion timely and report when congestion occurs. The congestion control Agent will select suitable congestion control algorithm to control congestion according to the reports from congestion inspection Agent. Using the collaboration of all agents to reduce the packet loss rate and increase network utilization.
Agent is an adaptive and intelligent software entity, which can represent user or other software to finish a task in a active service mode. It is autonomy, reactive, sociality, initiative, and adaptable. In this article, inspection Agent will inspect the congestion in the network and send reports, the control Agent will reduce or remove congestion, the yellow page Agent will provide service publish, service search, service subscription. The cooperation of all agents forms a multi-agent system providing services for networks. Based on the analysis and design of all agents’functions, the multi-agent system is developed for solving the network congestion using JADE platform.
引文
[1]刘勇生,魏亚东,袁淦泉.办公自动化实用教程[M].北京:电子工业出版社, 2006:3~11
[2]谢希仁编著.计算机网络[M].北京:电子工业出版社,2003.6:253~256
[3] Handley M,Floyd S, Padhye J, Widmer J, TCP Friendly Rate Control (TFRC): Protocol Specification.RFC 3448, January 2003
[4] Bansal Deepak, Balakrishman Hari. Binominal congestion control algorithm. In: IEEE INFOCOM 2001. Anchorage:AK, April 2001
[5] Yang Richard Yang, Min sik Kim, Simon S Lam. Analysis of Binomial congestion control. Technical Report TR-00-14, The university of Texas.Austin: June 2000
[6] Rejaie R,Handlery M, Estrin D, RAP: An End-to-End Rate-based Congestion Control Mechanism for Realtime Streams in the Internet. In: IEEE INFOCOM 1999, New York:1999
[7] Hewitt C. A universal modular actor formalism for AI. In Proceedings of the 3rd International Joint Conference on Artificial Intelligence (IJCAI-73), 1973:255~245
[8] Gasser L, Braganza C,Hermann N. MACE: a flexible test bed for distributed AI research. In Distributed Artificial Intelligence, 1987:119~152
[9] Wooldridge M, Jennings N R. Intelligence Agents: theory and practice. The Knowledge Engineering Review, 1995, 10(2): 115~152
[10] Wooldridge M, Jennings N R. The cooperative problem solving process. Journal of Logic and Computation, 1999, 9(4): 563~592
[11] Wooldridge M, Jennings N R. Pitfalls of Agent-oriented development. In Proceedings of the 2nd International Conference on Autonomous Agents (Agents 98), Minneapolis St. Paul, MN, 1998: 385~391
[12] Sycara K P. Multi-Agent compromise via negotiation. In Distributed Artificial Intelligence, VolumeⅡ, 1989: 119~138
[13] Davis R, Smith R J. Negotiation as a metaphor for distributed problem-solving. Artificial Intelligence. 1983, 20(1): 63~109
[14]廖勇,周德松,麻信洛等编著.流媒体技术入门与提高[M].北京:国防工业出版社, 2006.1:7~9
[15]鲁飞.视频实时传输拥塞控制机制的研究[D].武汉:中南民族大学, 2008
[16]张海军,杨印根,吴克捷等.基于RTP/RTCP协议的拥塞控制策略研究[J].电脑知识与技术,2008.12,Vol.4:2612~2613
[17]牛磊.网络拥塞问题研究及仿真[D].合肥:合肥工业大学,2008.4: 5-6.
[18] Low S H. TCP Congestion Controls: Algorithms and Models, Tutorial Slides, http://netlab.caltech.edu, 2000
[19]刘拥民,蒋新华,年晓红等. Internet端到端拥塞控制研究综述[J].计算机科学,Vol.35 No.2,2008: 7-9
[20] Wang H A, Schwartz M. Achieving Bounded Fairness for Multicast and TCP Traffic in the Internet. In Processing of ACM SIGOMN, 1998:81~82
[21]王小玲,梁兴建. Tahoe TCP拥塞控制算法研究和改进[J].四川理工学院学报,第19卷,第5期,2006年10月:62-64
[22]胡晓峰,孙志刚,苏金树.基于New Reno拥塞控制的TCP分组乱序影响分析[J].计算机工程与科学,Vol.31,No.5,2009:9-10
[23]杨彦彬. TCP拥塞控制机制浅析[J].通信技术,Vol.42,No.04,2009:58-59
[24]徐昌彪,鲜永菊编著.计算机网络中的拥塞控制和网络控制[M],人民邮电出版社, 2007.10:20-65
[25]黄辉春.网络拥塞控制技术研究及其仿真实现[D].哈尔滨工程大学,2005.1:25
[26]岳鹏.因特网拥塞控制机制若干问题的研究[D].西安电子科技大学,2006. 3:4-7
[27]刘靖,陈芳炯,韦岗.改进TCP Vegas拥塞控制协议及其在无线链路中的应用[J],科学技术与工程,Vol.7,No.3,Feb.2007: 339-340
[28] K. N.Srijith, Lillykutty Jacob, A.L. Ananda. TCP Vegas-A: Improving the performance of TCP Veags [J], Computing communications, 2005,28(4):429-440
[29]孟川杰.TCP Vegas的拥塞控制算法研究[D].四川大学,2006.544-545
[30]于斌,孙斌,温暖等编著. NS2与网络模拟[M].人民邮电出版社,2007:22-132.
[31]王辉编著. NS2网络模拟器的原理和应用[M].西北工业大学出版社,2008.8:20-123
[32] http://jade.cselt.it/
[33] Fabio Bellifemine,Giovanni Caire,Dominic Greenwood, Developing multi-agent systems with JADE, Wiley Series in Agent Technology, 2007:32~50
[34]张伟.基于Agent的局域网信息检索系统研究与实现[D].烟台:烟台大学,2009.3:37-45
[35] The Foundation for Intelligent Physical Agents. FIPA ACL Specification. http://www.fipa.org
[36] Fabio Bellifemine,Giovanni Caire,Dominic Greenwood, Developing multi-agent systems with JADE, Wiley Series in Agent Technology, 2007:100~107
[37]颉会芳.基于多Agent的土壤重金属污染预警模型研究[D].陕西:西北农林科技大学, 2010.5:22~24
[38]朱培红.基于移动多Agent的分布式网络性能检测的研究[D].武汉:武汉大学, 2004.5:52~55
[39]秦楠.基于多Agent系统的网络拥塞控制若干问题的研究[D].上海:同济大学, 2007.3,51~52
[40]张伟.基于Agent的局域网信息检索系统研究与实现[D].烟台:烟台大学,2009.3:37-45
[41]胡芳华,资武成.基于JADE平台的Multi-Agent协商的设计与实现[J],电脑知识与技术,2009.12, Vol.5, No.34:9695~9696
[42]王珊珊.基于JADE的多Agent系统仿真研究[D].上海:华中科技大学,2006.4:25~26
[2]谢希仁编著.计算机网络[M].北京:电子工业出版社,2003.6:253~256
[3] Handley M,Floyd S, Padhye J, Widmer J, TCP Friendly Rate Control (TFRC): Protocol Specification.RFC 3448, January 2003
[4] Bansal Deepak, Balakrishman Hari. Binominal congestion control algorithm. In: IEEE INFOCOM 2001. Anchorage:AK, April 2001
[5] Yang Richard Yang, Min sik Kim, Simon S Lam. Analysis of Binomial congestion control. Technical Report TR-00-14, The university of Texas.Austin: June 2000
[6] Rejaie R,Handlery M, Estrin D, RAP: An End-to-End Rate-based Congestion Control Mechanism for Realtime Streams in the Internet. In: IEEE INFOCOM 1999, New York:1999
[7] Hewitt C. A universal modular actor formalism for AI. In Proceedings of the 3rd International Joint Conference on Artificial Intelligence (IJCAI-73), 1973:255~245
[8] Gasser L, Braganza C,Hermann N. MACE: a flexible test bed for distributed AI research. In Distributed Artificial Intelligence, 1987:119~152
[9] Wooldridge M, Jennings N R. Intelligence Agents: theory and practice. The Knowledge Engineering Review, 1995, 10(2): 115~152
[10] Wooldridge M, Jennings N R. The cooperative problem solving process. Journal of Logic and Computation, 1999, 9(4): 563~592
[11] Wooldridge M, Jennings N R. Pitfalls of Agent-oriented development. In Proceedings of the 2nd International Conference on Autonomous Agents (Agents 98), Minneapolis St. Paul, MN, 1998: 385~391
[12] Sycara K P. Multi-Agent compromise via negotiation. In Distributed Artificial Intelligence, VolumeⅡ, 1989: 119~138
[13] Davis R, Smith R J. Negotiation as a metaphor for distributed problem-solving. Artificial Intelligence. 1983, 20(1): 63~109
[14]廖勇,周德松,麻信洛等编著.流媒体技术入门与提高[M].北京:国防工业出版社, 2006.1:7~9
[15]鲁飞.视频实时传输拥塞控制机制的研究[D].武汉:中南民族大学, 2008
[16]张海军,杨印根,吴克捷等.基于RTP/RTCP协议的拥塞控制策略研究[J].电脑知识与技术,2008.12,Vol.4:2612~2613
[17]牛磊.网络拥塞问题研究及仿真[D].合肥:合肥工业大学,2008.4: 5-6.
[18] Low S H. TCP Congestion Controls: Algorithms and Models, Tutorial Slides, http://netlab.caltech.edu, 2000
[19]刘拥民,蒋新华,年晓红等. Internet端到端拥塞控制研究综述[J].计算机科学,Vol.35 No.2,2008: 7-9
[20] Wang H A, Schwartz M. Achieving Bounded Fairness for Multicast and TCP Traffic in the Internet. In Processing of ACM SIGOMN, 1998:81~82
[21]王小玲,梁兴建. Tahoe TCP拥塞控制算法研究和改进[J].四川理工学院学报,第19卷,第5期,2006年10月:62-64
[22]胡晓峰,孙志刚,苏金树.基于New Reno拥塞控制的TCP分组乱序影响分析[J].计算机工程与科学,Vol.31,No.5,2009:9-10
[23]杨彦彬. TCP拥塞控制机制浅析[J].通信技术,Vol.42,No.04,2009:58-59
[24]徐昌彪,鲜永菊编著.计算机网络中的拥塞控制和网络控制[M],人民邮电出版社, 2007.10:20-65
[25]黄辉春.网络拥塞控制技术研究及其仿真实现[D].哈尔滨工程大学,2005.1:25
[26]岳鹏.因特网拥塞控制机制若干问题的研究[D].西安电子科技大学,2006. 3:4-7
[27]刘靖,陈芳炯,韦岗.改进TCP Vegas拥塞控制协议及其在无线链路中的应用[J],科学技术与工程,Vol.7,No.3,Feb.2007: 339-340
[28] K. N.Srijith, Lillykutty Jacob, A.L. Ananda. TCP Vegas-A: Improving the performance of TCP Veags [J], Computing communications, 2005,28(4):429-440
[29]孟川杰.TCP Vegas的拥塞控制算法研究[D].四川大学,2006.544-545
[30]于斌,孙斌,温暖等编著. NS2与网络模拟[M].人民邮电出版社,2007:22-132.
[31]王辉编著. NS2网络模拟器的原理和应用[M].西北工业大学出版社,2008.8:20-123
[32] http://jade.cselt.it/
[33] Fabio Bellifemine,Giovanni Caire,Dominic Greenwood, Developing multi-agent systems with JADE, Wiley Series in Agent Technology, 2007:32~50
[34]张伟.基于Agent的局域网信息检索系统研究与实现[D].烟台:烟台大学,2009.3:37-45
[35] The Foundation for Intelligent Physical Agents. FIPA ACL Specification. http://www.fipa.org
[36] Fabio Bellifemine,Giovanni Caire,Dominic Greenwood, Developing multi-agent systems with JADE, Wiley Series in Agent Technology, 2007:100~107
[37]颉会芳.基于多Agent的土壤重金属污染预警模型研究[D].陕西:西北农林科技大学, 2010.5:22~24
[38]朱培红.基于移动多Agent的分布式网络性能检测的研究[D].武汉:武汉大学, 2004.5:52~55
[39]秦楠.基于多Agent系统的网络拥塞控制若干问题的研究[D].上海:同济大学, 2007.3,51~52
[40]张伟.基于Agent的局域网信息检索系统研究与实现[D].烟台:烟台大学,2009.3:37-45
[41]胡芳华,资武成.基于JADE平台的Multi-Agent协商的设计与实现[J],电脑知识与技术,2009.12, Vol.5, No.34:9695~9696
[42]王珊珊.基于JADE的多Agent系统仿真研究[D].上海:华中科技大学,2006.4:25~26