On the use of classical control system based AQM for rate adaptive streaming media

摘要: The Internet is today a world wide packet switching arena constituting enormous possibilities of new services and business creation. E.g., there clear tendency that more real-time are making the jump from dedicated circuit-switched broadcasting networks into switching. Examples telephony, videoconferencing, television. thus hosting large set different services, including delay tolerant Web-surfing traffic, but also non-delay services. An additional challenge with most traffic its pattern do not adapt to varying load as Web-traffic do. Still, these work well, long capacity sufficient. Problems arise when growth service usage larger than increase. During peak hours, users will then start experience media fall-out excessive communication delay.The reason we know it was built handle such at all. In motor comparison, queues build up road crossover capacity. behaves in similar fashion: information sent packets can be compared cars. If too many heading same direction, routers, causing extra during hours. one way fearful traffic: if get long, arriving simply dropped, i.e. they just vanish. Luckily, no direct parallel this phenomenon comparison realm! To assist queuing problems special lanes defined allow e.g. only buses, taxis, cars where driver has least passenger, drive lane.Thus, road-users less hours rest population. tried “healed” some comparable means. E.g. use IntServ or DiffServ Quality Service, belonging high priority applications treated preferential fashion. But what happens “special-lanes”? What total over-loaded over significant time period? answer fix problem simple: aggregate generation must slow down! means each car carries people (i.e. fewer total), or, equivalently, big exchanged by smaller cars, producing queues. multimedia realm, equivalent solution content compressed efficiently, and/or packets.This thesis proposes for live interactive streaming tight interaction between sources network very essential. A novel router architecture, “P-AQM”, switched core component. primary P-AQM design objective native support rate adaptive flows, addressing low losses even conversational flows. second bandwidth fairness among elastic (TCP) These two objectives achieved due aforementioned routers sources: signal congestion level, while TCP apply adaption. built-in control mechanism (e.g. Tahoe Reno) reacts on drops ECN tags performed router. Real-time using UDP protocol standardized mechanism. While DCCP/TFRC become compelling IETF standard last years, chosen another bypasses TFRC performance. Using level signals done much precise, react faster changes, obtain intra-flow global max-min fairness. cost improvements gradual deployment added signaling traffic.The following classical theory principles, been developed improved combination analytical simulation tools. As side effect need true decodable video framework tool-set, “Evalvid-RA”, generate traffic. Evalvid-RA other researchers improving their own work, applying codecs protocol.