Development of a Content Delivery Network for Next Generation and its Performance Analysis

点击翻译按钮获取中文摘要

We have carried out our research actively towards developing a highly efficient content delivery network from theoretical point of view. Some of our achievements are listed below.・We have investigated the impact of long-range dependence arising in traffic on the performance of real-time content delivery.・We have proposed a channel allocation algorithm for a cellular network which allocates channel bandwidth efficiently giving higher priority to voice calls and derived performance measures such as packet loss probability, transmission delay and call loss probability.・We have developed a mathematical model called a 'polling model' to evaluate the performance of Bluetooth technology for the last hop of content delivery and analytically derived the transmission delay and throughput of the model.・We have analyzed the performance of some variants of FEC which are based on XOR and clarified the relationship among 1)the number of symbols taken together in a XOR operation, 2)its probability distribution if it is random, 3)the ratio of redundancy to the original data set, 4)the channel error rate 5)the successful transmission probability.・We have considered the performance of a hybrid FEC/ARQ for the recovery from packet losses and bit errors in real-time content distribution.・We have proposed an autonomous self-organization algorithm for distributed content delivery systems using overlay networks and demonstrated its usefulness.・We have developed a mathematical model and its performance analysis for the appropriate design and dimensioning of playout delay and buffer size taking network delay caused by congestion into consideration.・We have developed a fundamental technique to estimate the traffic volume and distributions of hop counts and round trip time between peers in overlay networks and carried out experimental tests in real networks.