A Framework for Fast Congestion Detection in Wireless Sensor Networks Using Clustering and Petri Net-based Verification Khanh Le 1 , Thang Bui 1 , Tho Quan 1 , and Laure Petrucci 2 1 Ho Chi Minh City University of Technology, Vietnam {lnkkhanh,thang,qttho}@cse.hcmut.edu.vn 2 LIPN, CNRS UMR 7030, Universit´ e Paris 13, Sorbonne Paris Cit´ e, France Laure.Petrucci@lipn.univ-paris13.fr Abstract. Applications of Wireless Sensor Networks (WSN) in harsh conditions usually cover a vast area with sensors randomly deployed by an uncontrolled method, e.g. dropped by helicopters. Thus the actual topology is unpredictable and can suffer from possible congestion. We propose the FCD framework for congestion detection, based on clustering techniques combined with Petri nets modelling and verification. 1 Introduction The congestion problem in Wireless Sensor Networks (WSNs): A WSN is a collection of hundreds or thousands of sensors. Sensors are cheap, low energy consuming devices, with limited memory and processing capabilities [1]. They can communicate with one another using WiFi. Depending on the targeted application, a WSN is deployed in a dense or sparse mode. Environment moni- toring WSNs are usually implemented on a dense network topology [6] whereas some applications require sparsely spreading sensors over a large geographical area, e.g. for tracking transportation in a city [8]. The limited processing capac- ity and energy of sensors has some disadvantages, hence the necessity of some QoS (Quality of Service) constraints such as delay, security or congestion. We here focus on congestion detection. Following [11], congestion depends on the network topology and the trans- mission rate. Indeed, sensors in a dense network are deployed at very close dis- tances, thus some packets transmitted over the same paths may collide. Also, the processing rate of sensors may be smaller than their receiving rate: if the transmission rate is too high, the sensor’s buffer overload can induce congestion. Similarly, the congestion also occurs during transmission in sparse deployment networks if a huge number of packets are sent within a short time. Packets loss and retransmission may cause the overload at some sensors. Tools for Congestion Detection: The main approaches to congestion de- tection are simulation or model based. In the first case, a simulator is used to mimic the operations of the WSN, measure the performance, and check whether a certain anomaly like congestion occurs or not. Widely used simulators include