Author Description

The Internet has revolutionized the way information is shared between geographically separated entities and has become the medium of choice for data sharing for both commercial and military units. This reality seemingly simplifies network design, but characteristics of the underlying transport protocols require consideration when used in a high delay, high error environment, such as a satellite transmission architecture. Of great interest are modifications to the Transmission Control Protocol (TCP) that enhance TCP performance in these poor conditions. Although extensive research has been conducted concerning TCP optimization, few have assessed the performance benefit a given modification can provide. The purpose of this study was to assess the throughput improvement afforded by the various TCP optimization techniques, with respect to a simulated geosynchronous satellite system, in order to provide a cost justification for the implementation of a given enhancement technique. It was determined that each technique studied, which included the Space Communication Protocol Standard - Transport Protocol (SCPS-TP), window scale, selective acknowledgements (SACKs), and combinational use of the window scale and SACK mechanisms, provided varying levels of improvement as compared to a standard TCP implementation. In terms of throughput, SCPS-TP provided the greatest overall improvement, with window scale and window scale/SACK techniques providing significant benefits at low levels of bit error rate (BER). The SACK modification improved throughput performance at high levels of BER, but performed at levels comparable to standard TCP during scenarios with lower BER levels.