Modeling local broadcast behavior of wireless sensor networks with timed automata for model checking of WCTT

Wireless Sensor Networks (WSNs) are usually deployed in order to monitor parameters of an area. When an event occurs in the network an alarm is sent to a special node called the sink. In critical real-time applications, when an event is detected, the Worst Case Traversal Time (WCTT) of the message must be bounded. Although real-time protocols for WSNs have been proposed, they are rarely formally verified. The model checking of WSNs is a challenging problem for several reasons. First, WSNs are usually large scale so it induces state space explosion during the verification. Moreover, wireless communications produce a local broadcast behavior which means that a packet is received only by nodes which are in the radio range of the sender. Finally, the radio link is probabilistic. The modeling of those aspects of the wireless link is not straightforward and it has to be done in a way that mitigate the state space explosion problem. In this paper we particularly focus on the modeling of the local broadcast behavior with Timed Automata (TA). We use TA because they have sufficient expressiveness and analysis power in order to check time properties of protocols, as shown in the paper. Three ways of modeling local broadcast with synchronizations of TA are presented. We compare them and show that they produce different state space sizes and execution times during the model checking process. We run several model checking on a simple WSN protocol and we conclude that one model mitigate the state explosion problem better than the others. In the future, the next step will be to enhance this model with the probabilistic aspect of radio communications and to show it remains the best one.