Research On Media Stream Rate Distribution Strategy

With wide band technology developing, stream media application increases rapidly and the capacity of media stream on Internet rises like an arrow. But network bandwidth is always limited, for the purpose of high network QoS of media steams, all media streams must be kept under control. In this thesis, from the angle of network operator's point and system viewpoint, researches on streams'rate distribution on large capacity network have been done. Media related technology is studied through network model for large quantity of media streams on large network, Optimal Rate Distributed Strategy(ORDS), Optimal Access Control Strategy(OACS), two kinds of distributed algorithm(CRDA and CDDA) and Multiple Asymptotic Optimal Stream Rate Distribution Strategy (MARDS) are researched thoroughly. For the more, media network QoS are analyzed for storied media and live media, and so do link used rate and network blocking rate. At last, simulations show that media streams should be assigned different rate and only in this way network QoS can be upgraded.Main researches can be summarized as four aspects below:1. Optimal Rate Distribution Strategy (ORDS):Stored rate-adaptive media stream is researched and so do live rate-adaptive media stream too. On single link for the stored rate-adaptive media stream, the full expression of the expected time-averaged rate level is deduced thoroughly under ORDS. Then asymptotic optimal rate distribution strategy is studied with just two rate level and research results show that rate distribution strategy with two media rate level can have the same QoS as that under ORDS.2. Optimal access Control Strategy (OACS):Under this strategy, the control threshold capacity and media QoS are investigated thoroughly. And simulation result shows that the expected time-averaged rate level is nearly the same as the value under ORDS. 3. Distributed algorithm (CRDA and CDDA):Two kinds of distributed algorithm are in details discussed. CRDA is designed according to the theory ORDS and OACS and compared with CDDA. The simulation result shows that the expected time-averaged rate level of network is higher with CRDA between two distributed algorithm and the asymptotic expected time-averaged rate level value is almost equal under the capacity dependent algorithm and optimal adaptation policy.4. Multiple Asymptotic Optimal Stream Rate Distribution Strategy (MARDS): MARDS are investigated under common network and single link. The close expression for the asymptotic expected time-averaged rate level is investigated under the multiple services network also. In the last chapter, as an example multiplexing audio and video streams are discussed.