The Transmission Control Protocol has been designed to support interactive and bulk applications, with per-formance tuned to support bulk applications that desire to continuously send data. In contrast, this paper analyses TCP performance for a class of applications that do not wish to send continuous data, but instead generate bursts of data separated by application-limited periods in which little or no data is sent. In this con-text, the paper evaluates an experimental method, Congestion Window Validation (CWV), proposed to miti-gate the network impact of bursty TCP applications. Simulation results show that TCP-CWV exhibits a con-servative behaviour during application-limited periods. The results also show that TCP-CWV is able to use the available capacity after an idle period over a shared path and that this can have benefit, especially over long delay paths, when compared to slow-start restart specified by standard TCP. The paper recommends the development of CWV-like algorithms to improve the performance for bursty applications while also pro-viding an incentive for application designers to use congestion control.
|Number of pages||7|
|Journal||International Journal of Communications, Network and System Sciences|
|Publication status||Published - 1 Jul 2010|
- transport layer