SACK TCP resilience

The selective-acknowledgement transmission control protocol (SACK TCP) is the most up-to-date version of TCP commonly used in the Internet. Understanding the resilience behaviour of SACK TCP is very important because congestion episodes and failure events are relatively frequent. In this paper, the descriptive power of Petri nets is leveraged to present the reactions that take place inside SACK TCP when network failures or congestion occur. An algebraic model is also presented that can be used to calculate the critical number of lost segments, nC, above which SACK TCP resilience is significantly degraded.