On the delay distribution and maximum message length in DTNs with long propagation delays

We propose a model for computing the optimal message length in delay tolerant networks with long propagation delays. As a protocol of interest we use CCSDS File Delivery Protocol operating in differed NAK mode. The proposed model is based on the absorbing Markov chain framework allowing to get not only the mean performance metrics but the probability mass function of time it takes to transmit a message. The latter is due to the special structure of the absorbing Markov chain allowing for simple estimation of the first passage time till absorbtion. Using the pmf one can determine the maximum message length that can be delivered with a certain reliability when two nodes remain in coverage range of each other for a given time.