An experimental evaluation of a novel acknowledgment scheme over GEO-satellite links

Selective acknowledgement (SACK) and negative acknowledgment (NAK) are well-known schemes in the current Internet. Selective negative acknowledgement (SNACK) is a novel scheme which integrates the capabilities of both SACK and NAK, developed for space Internet. There has been some work done in discussing the performance of the SNACK scheme based on the methodology of the scheme or by considering SNACK as an isolated entity. It is necessary to have a side-by-side, controlled experimental comparison of the effectiveness of SNACK and the widely used SACK with the operating systems fully involved. This paper evaluates the effectiveness of SNACK over GEO-satellite links in an experimental manner using the SGLS testbed, by comparing it with the widely used SACK. It has been found that at low BERs (<=10/sup -6/), SNACK and SACK show similar performance. At a higher BER (around 10/sup -5/), SNACK significantly increases link utilization and shows superior performance over SACK. SNACK is better suited for long delay and high BER channels and thus makes it favorable over SACK in the experimented communication environments.