Improving TCP performance after a long channel outage
H. Murakami, Gang Wu, Masugi Inoue · 2003
Bursty packet loss caused by channel outage is one of the most significant problems in high-frequency-band communications because TCP applications, especially after experiencing bursty packet loss, take a long time to recover. This paper proposes a rapid recovery algorithm for use at the TCP layer for a bursty packet loss channel in a server-client model. The algorithm, which needs to be installed only in a client, includes transmission of "pseudo acknowledgement packets" and adjustment of advertised receive window size as soon as the loss ends. The algorithm works with a standard TCP-Reno version with a SACK (selective acknowledgement) option used in both the server and the client. For instance, for a 0.5-second-long outage, the standard TCP takes at least a few minutes to recover because of the native slow-start algorithm, while the TCP with the proposed algorithm can resume the transmission immediately after the client learns that the channel has recovered.