Packet delay modeling of truncated multi-process ARQ protocols for parallel communications

This paper presents a packet transmission delay modeling for networks using multiple truncated stop-and-wait (SAW) retransmission processes. The packet transmission delay includes the packet queueing delay (PQD) and the packet reordering delay (PRD). While the first type of delay is commonly known for any SAW procedure, the latter is only introduced when multiple SAW processes are considered for parallel communications. These processes work independently while sharing the same packet traffic. By modeling the queueing buffer as an M/G/∞ model, an analytical formula for the average PQD is provided. In addition, we model the PRD as a first-order Markov chain defined by a transition probability matrix, and derive the average PRD. Analytical formulae are provided considering the general case where error probabilities are different from a transmission to another, such as in automatic repeat request (ARQ) with packet combining, and for arbitrary values of the number of multi-SAW processes and maximum number of ARQ retransmissions.