| Applications like IPTV, online meeting and remote education, which use live streaming as core technology, are getting more and more popular, and the scale of live events are also growing rapidly.As the scale of live streaming getting bigger and bigger, service provider’s exit bandwidth increases linearly with receiver numbers, and therefor costs a lot. Using CDN and P2P technology can solve the bandwidth problem. But CDN makes server system difficult to build, deploy and maintain, and leads to low universality. Live streaming using P2P results in large delay and out of sync at clients.The best known method to do live streaming is multicast, which leads to good bandwidth efficiency. But as many routers in the Internet backbone do not support or disable multicast, multicast technology is not widely applied.As the existing technologies don’t perform well at low-delay, bandwidth and universality, in this essay we propose a streaming solution using native multicast and Automatic Multicast Tunneling (AMT). This paper focuses on the following problems and proposes solutions:1. Waiting time caused by reference frame missing. This essay sets the maximum reference frame interval to confine waiting time, and quantifies its relation with waiting time and stream bitrate.2. Auto bitrate adaption on receiver side. This essay uses receiver driven adjust strategy, and conservative bandwidth assessment strategy.3. Transportation of media from application sharing. This essay studies bitrate and content feature, uses leaky bucket in bitrate adaption.By including AMT, multicast can cover more end-users. The solution can solve the bandwidth problem in large scale multimedia streaming, at the same time providing acceptable user experience. |
