Modeling end-to-end reliability in best-effort networked embedded systems

In the networked control systems with the best-effort message delivery, there is no guarantee that the application data are delivered to the destination(s) successfully. To improve transmission reliability in such systems, the messages with application data are sent usually more than once. In the unacknowledged repeated service, the number of message repetitions is fixed a priori. In the acknowledged service, the messages are retransmitted until the sender receives an acknowledgment of a successful message reception. The number of repetitions in the former, and the maximum number of retries in the latter are configured by the user. The contribution of the paper is the analytical modelling of the probabilities that the transactions of various types are completed successfully.