Evaluation of the TCP TS-Prio protocol using Markovian models

The quality of service (QoS) perceived by end users in their access to communication services is mainly driven by transport level protocols, like TCP. Recently a protocol that implements service differentiation by appropriate modification of the sliding window size considering the data transfer priority, called TCP TS-Prio (Transient & Static Priority) protocol, was proposed. In this paper, we propose a methodology that can be used to estimate key performance parameters, such as packet loss, queueing delays, window sizes, and throughputs. Our approach models a single TCP TS-Prio source using a Markovian chain. The superposition and interaction of several TCP sources, with service differentiation, is then analyzed. The analysis shows that TCP TS-Prio is able to offer service differentiation depending on the network congestion.