Détection et diagnostic des fautes dans des systèmes à base de réseaux de capteurs sans fils

Abstract : Sensor faults are the rule and not the exception in every Wireless Sensor Network (WSN) deployment. Sensor nodes are fragile, and they may fail due to depletion of batteries or destruction by an external event. In addition, nodes may capture and communicate incorrect readings because of environmental influence on their sensing components. Links are also failure-prone, causing network partitions and dynamic changes in network topology, leading to delays in data communications. Links may fail when permanently or temporarily blocked by an external or environmental condition. Packets may be corrupted due to the erroneous nature of communications. When nodes are embedded or carried by mobile objects, nodes can be taken out of the range of communications. WSNs are also prone to malicious attacks, such as denial of service, injection of faulty packets, leading to unexpected behavior of the system and so on. In addition to these predefined faults or failures (i.e., with known types and symptoms), many times the sensor networks exhibits silent failures that are unknown beforehand and highly system-related. Applications over WSNs, in particular safety critical applications, such as fire detection or burglar alarm systems, require continuous and reliable operation of the system. However, validating that a WSN system will function correctly at run time is a hard problem. This is due to the numerous faults that can be encountered in the resource constrained nature of sensor platforms together with the unreliability of the wireless links networks. A holistic fault management approach that addresses all fault issues does not exist. Existing work most likely misses some potential causes of system failures. The reason is simple : the more elements to monitor, the more information to be collected and sometimes to be exchanged, then the more the energy consumption becomes higher. In this thesis, we propose an Integrated Fault Tolerance Framework (IFTF) that provides a complete picture of the system health with possibility to zoom in on the fault reasons of abnormal phenomena. IFTF detects data anomalies, diagnoses network failures, detects application level failures, identifies affected areas of the network and may determine the root causes of application malfunctioning. These goals are achieved efficiently through combining a network diagnosis service (component/element level monitoring) with an application testing service (system level monitoring) and a data validation system. The first two services reside on each node in the network and the data validation system resides on each cluster head. Thanks to IFTF, the maintenance and reconfiguration operations will be more efficient leading to a more dependable WSN. From the design view, IFTF offers to the application many tunable parameters that make it suitable for various application needs. Simulation results show that the presented solution is efficient both in terms of memory use and power consumption. Data validation system does not incur power consumption (communication overhead). Using testing service combined to diagnosis service incurs a 4 %, on average, increase in power consumption compared to using solely network diagnosis solutions.
Document type :
Complete list of metadatas

Cited literature [138 references]  Display  Hide  Download

Contributor : Abes Star <>
Submitted on : Friday, February 28, 2014 - 11:02:24 AM
Last modification on : Thursday, June 21, 2018 - 11:44:00 AM
Long-term archiving on : Wednesday, May 28, 2014 - 11:06:09 AM


Version validated by the jury (STAR)


  • HAL Id : tel-00953241, version 1




Dima Hamdan. Détection et diagnostic des fautes dans des systèmes à base de réseaux de capteurs sans fils. Autre [cs.OH]. Université de Grenoble, 2013. Français. ⟨NNT : 2013GRENM007⟩. ⟨tel-00953241⟩



Record views


Files downloads