Estimation of the advanced TCP/IP algorithms for long distance collaboration

ITER collaboration needs the wide bandwidth network for the fast data transfer. However, the throughput might be too small to transfer the bulk data, because the delay of the signal is not negligible. This problem comes from the mechanism of the communication protocol, TCP/IP. In this paper, long fat pipe network (LFN) environment was emulated by the Linux-based PC. The effects of the various congestion avoidance algorithms were investigated with the delay which was estimated between ITER site in France and ITER satellite site in Japan. The standard algorithm, Reno, is not suitable for LFN. The advanced algorithm, CUBIC-TCP, Hamilton TCP and Scalable TCP can be more suitable for LFN such as the ITER international collaboration.