Intermittent communication and partial divergence

We formulate a model for intermittent communication in which the transmissions of information symbols are bursty or the channel is sporadically available. We consider a general scenario in which information and noise symbols are grouped into packets of length l, and noise packets are inserted in the input sequence of the channel, and the receiver does not know a priori the positions of the information packets. Depending on the scaling behavior of the packet length relative to the codeword length, we identify some interesting scenarios for the scaling behavior of the receive window relative to the codeword length, and find achievable rates using different decoding structures. Interestingly, some of the decoding structures are based on a generalization of the method of types and properties of partial divergence. The achievable rates and the numerical results confirm the intuitive idea that increasing the intermittency decreases the achievable rates by increasing the uncertainty about the positions of the information packets at the receiver.