Anomaly Detection with Wireless Sensor Networks. Nathalie Dessart ∗ , Hac` ene Fouchal † , Philippe Hunel ∗ and Nicolas Vidot ∗ ∗ LAMIA Universit´ e des Antilles et de la Guyane, France Email: ndessart, phunel, nvidot@martinique.univ-ag.fr † CReSTIC, Universit´ e de Reims Champagne-Ardenne Email: Hacene.Fouchal@univ-reims.fr Abstract—The aim of this study is to suggest two automated techniques able to help medical staff to detect earlier than usual some diseases using wireless sensor networks (WSNs). In this context, a patient is equipped with physical sensors which sense health parameters. This WSN will perform some computations and will run an alarm when a disease is suspected. The first technique uses a population protocol to handle data exchanged between motes and provides an efficient algorithm to suggest that a disease is diagnosed on a patient. The algorithm is distributed, i.e., the decision may be done by any sensor dealing with the disease detection. The second technique uses a token algorithm where, some motes are denoted as masters. Each of them is in charge of deciding if a specific disease occurs. This technique is not totally distributed but enhances the network efficiency regarding to the energy consumption, the time execution and the number of exchanged messages. Keywords-Wireless sensor networks; Distributed decision; Population protocols; I. I NTRODUCTION Wireless sensor networks can be used in a medical context to monitor patients. In such a context, a patient is linked to one or more motes (a sensor equipped with transducers) and each mote measures a health parameter (heart rate, electrocardiogram, temperature, blood sugar, ...). The raw data can be then stored on a centralized server to be analyzed and interpreted. There exist medical equipments which allow doctors to access to the patients medical history in a wireless manner. Some of them use bluetooth [3][2], and others work on the Wireless Medical Telemetry (WMTS) bands [4]. In this study, we intend to use WSNs in order to provide better monitoring, to perform some data filtering on motes and to enhance patient comfort (less wired equipments). Better monitoring would be provided since tasks will be decentralized (computation is not done by a single engine) and mobility will be provided, allowing doctors and nurses to require patients data via a mobile equipment or patients to move easily. This mobility can also be useful in an intensive care context when a patient has to pass some medical exams. One of the most important challenges in the medical context is to assist doctors to diagnose diseases as soon as possible. A disease may be diagnosed once some medical exams have been performed on a patient (X-rays, blood analysis, ...) or by monitoring some health parameters with simple transducers. In the last case, a disease can be diagnosed if some parameters have reached some specific thresholds. In order to establish a diagnosis, medical rules have been used since decades. Some of these rules can be expressed as inequalities and thus if an inequality is verified a disease is suspected. We propose to equip each sensor with one transducer. Each sensor will compare a sensed health parameter and its threshold. Then, the collaboration between sensors, will compute the diagnosis and an alarm will be raised whether a disease is suspected. We studied two approaches to implement this strategy. The first one uses the population protocol concept to compute the collaboration between motes. The second one will use the token concept which will allow a specific mote to collect all data related to one disease. Thus, this mote will be able to diagnose the given disease. This paper is organized as follow, section 2 discusses briefly related works to medical monitoring. In section 3, we draw our context and introduce some definitions. We present a centralized diagnosis algorithm which will be the comparison basis n section 4. Section 5, details the use of both population protocol and token algorithm. In the last section, we conclude and draw some perspectives. II. RELATED WORK Several works deal with monitoring patients using WSN. In [5], authors proposed a software infrastructure (named Codeblue) for emergency medical care to be deployed at hospital or over an emergency site, as well as a hardware framework. They suggest a public subscribe architecture, in a TinyOs environment, written in nesC. They show that using WSNs for medical care is suitable enough. Authors of [6] focus on home assistance for elder people who live alone and propose SAILNet. Their main 2010 Ninth IEEE International Symposium on Network Computing and Applications 978-0-7695-4118-1/10 $26.00 © 2010 IEEE DOI 10.1109/NCA.2010.36 204