A Study On TFRC Based Video Transmission Evaluation And TFRC Optimization

As the existing Internet architecture is a model of the five layer sandglasses, alterations to the middle layers are now limited to simple incremental updates; What’s more, ISPs pursuit an ultimate ROI, which makes deployment of any new, radically different technology is nearly impossible because the investment returns only when the most of the Autonomous Systems have employed the same new technology. The two has caused network ossifications, which made the fact that RSVP, IPv6, Multicast etc have not been widely adopted.To address this ossification, network virtualization has been propounded as a diversifying attribute of the future inter-networking paradigm. We can construct multiple heterogeneous virtual networks from the Internet with network virtualization technology. We can test and deploy new network architectures in these virtual networks without introducing much impact to the Internet. What’s more, along with the prevailing of media applications, such as IP telephony and multimedia streams, higher requirements on service and resources have been imposed upon the underlying network. By means of virtual network construction, network virtualization techniques can be used to provide multi-media operation with greatly improved quality. As one of the transportation protocol proposed by ITEF for multi-media services, TFRC have received extensive attention. This paper investigated evaluation techniques for video transmission quality based on TFRC, and improved TFRC performance in connection with the situation when different rate of TFRC flows complete with each other.This paper suggests a simple frame for the evaluation of real-time video transmission on the top of SVEF. A dedicated buffer has been built to provide a higher transmit efficiency, which is next to real application. Through simulations under different bandwidth, it has proved that TFRC is more suitable than TCP.When conducting the above work, we found a fairness issue introduced by competing TFRC flows that have different bandwidth limits. We analyzed the cause, and found that the method TFRC employed to calculate loss event rate p is the major contributor. We proposed a virtual flow method to deal with the problem. It divided the relatively high rate flow to several virtual small flows, and calculates their throughput respectively. The sum of virtual flow throughputs is then used to evaluate the corresponding loss event rate. Through theoretical reasoning and simulations under various scenarios, this paper had proved that the virtual flow method can effectively improve fairness usage of shared bandwidth resources among different rates of TFRC flows.