Performance Analysis of a Network of Event-based Systems (1401.4935v1)

Abstract: We consider a scenario where multiple event-based systems use a wireless network to communicate with their respective controllers. These systems use a contention resolution mechanism (CRM) to arbitrate access to the network. We present a Markov model for the network interactions between the event-based systems. Using this model, we obtain an analytical expression for the reliability, or the probability of successfully transmitting a packet, in this network. There are two important aspects to our model. Firstly, our model captures the joint interactions of the event-triggering policy and the CRM. This is required because event-triggering policies typically adapt to the CRM outcome. Secondly, the model is obtained by decoupling interactions between the different systems in the network, drawing inspiration from Bianchi's analysis of IEEE 802.11. This is required because the network interactions introduce a correlation between the system variables. We present Monte-Carlo simulations that validate our model under various network configurations, and verify our performance analysis as well.