No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Analysis of Threshold-Based Event Detection Algorithms for Wireless Sensor Networks by Fault Injection
Abstract
Wireless sensor networks are often deployed in harsh environments where they are exposed to extreme conditions. The influences and faults resulting from these conditions are often overlooked. As there are many possibilities for the occurrence of faults, dependability considerations are of high importance. In this work we present an approach for choosing the most adequate parameter set out of a number of qualified parameter sets for an event detection algorithm (EDA). Instead of only considering the performance of the EDA for data that does not contain errors, we analyze the EDA's reaction to erroneous data by injecting different kinds of faults. In a case study, we show how a seemingly optimal parameter set becomes apparent to be having a weak point. We also show how an application engineer can be provided with feedback about weaknesses of the employed EDA and its parameter set, respectively.
ResearchGate has not been able to resolve any citations for this publication.
We report on preliminary experiences with deploying a large-scale sensor network (about 100 nodes) for a pilot in precision agriculture. The pilot did not answer the initial research questions, but instead revealed many engineering problems typically overlooked by (computer) scientists evaluating their work by means of simulation. The deployment prompted us to rethink our development process and includes important lessons for the WSN research community as a whole
This paper presents an effective approach for injecting faults/errors in WSN nodes operating systems. The approach is based on the injection of faults at the assembly level. Results show that depending on the concurrency model and on the memory management, the operating systems react to injected errors differently, indicating that fault containment strategies and hang-checking assertions should be implemented to avoid spreading and activations of errors.
Recently, wireless sensor networks (WSNs) have become mature enough to go beyond being simple fine-grained continuous monitoring platforms and become one of the enabling technologies for disaster early-warning systems. Event detection functionality of WSNs can be of great help and importance for (near) real-time detection of, for example, meteorological natural hazards and wild and residential fires. From the data-mining perspective, many real world events exhibit specific patterns, which can be detected by applying machine learning (ML) techniques. In this paper, we introduce ML techniques for distributed event detection in WSNs and evaluate their performance and applicability for early detection of disasters, specifically residential fires. To this end, we present a distributed event detection approach incorporating a novel reputation-based voting and the decision tree and evaluate its performance in terms of detection accuracy and time complexity.
The successful deployment of a wireless sensor network is a difficult task, littered with traps and pitfalls. Even a functional network does not guarantee gathering meaningful data. In SensorScope, with its high-mountain campaigns, we have acquired invaluable experience in planning, conducting, and managing real-life sensor network deployments. In this paper, we share our knowledge by stepping through the entire process, from the preparatory hard- and software development to the actual field deployment. Illustrated by numerous real-life examples, excerpted from our own experience, we point out many potential problems along this way and their possible solutions.
We propose a distributed solution for a canonical task in wireless sensor networks - the binary detection of interesting environmental events. We explicitly take into account the possibility of sensor measurement faults and develop a distributed Bayesian algorithm for detecting and correcting such faults. Theoretical analysis and simulation results show that 85-95 percent of faults can be corrected using this algorithm, even when as many as 10 percent of the nodes are faulty.
Teaching wireless sensor networks (WSNs) only theoretically is not sufficient to understand the complex interaction of these networks. WSNs consist of sensor nodes which measure physical quantities of their environment, preprocess the measured data, and transmit it towards a base station in a multi-hop manner. WSNs are typically used in application areas without wired infrastructure and so they must be powered by batteries or energy harvesting systems. Due to the influence of different factors on the behavior, practical exercises can enhance the learning process because the students can perform their experimentation independently. This work presents the use of a remote lab for teaching energy harvesting enhanced WSNs. Students can learn the behavior of WSNs and the influences of energy harvesting. Furthermore, practical aspects of WSNs are shown by using a realistic application scenario. This work is part of the European project Remotelabs Access in Internet-based Performance-centered Learning Environment for Curriculum Support.
Wireless sensor networks are being considered for use in industrial process and control environments. Unlike traditional deployment scenarios for sensor networks, in which energy preservation is the main design principle, industrial environments stress worker safety and uninterrupted production. To fulfill these requirements, sensor networks must be able to provide performance guarantees for radio communication. In this paper, we consider as a case study the deployment of a sensornet in an oil refinery in Portugal, where sensor nodes are deployed outdoors and might experience high temperature fluctuations. We investigate how the variations of ambient temperature influence data delivery performance and link quality in low-power radio communications. We also study the impact that specific implementation requirements, such as the ATEX fire-safety regulations, can have on the design of the overall network. Our experiments show that temperature directly affects the communication between sensor nodes, and that significantly less transmission power is required at low temperatures. We further illustrate that it is possible to save up to 16% energy during nights and cold periods of the year, while still ensuring reliable communication among sensor nodes. In view of these experimental results, we elaborate on how the temperature influences both the design and the deployment of wireless sensor networks in industrial environments.
Event detection using wireless sensor networks (WSNs) has become a new field of research in the past years, increasing the need for dependability and fault tolerance. Our work exploits the massive redundancy of large WSNs in combination with neighbours' relations to identify faulty nodes. We present a new approach to categorize nodes in being faulty or fault free based on the event detection results of the nodes' neighbours and the nodes adjacent to the neighbours. For error probabilities < 0.2 our algorithm performs closely to other work in the field, and performs considerably better for error probabilities up to 0.5.
Wireless Sensor Networks for surveillance systems in home, office, or factory environment require correct tracking of intruders. For such systems, passive infrared motion sensors (PIR sensors) are ideal because they do not require any signal or devices on the object to be tracked and they can work in dark environment as well. This paper first analyzes the performance and the applicability of the PIR sensors for security systems. Then, we propose a region-based human tracking algorithm with actual implementation and experiment in real environment. From the experiments, we show that the human tracking algorithm based on the PIR sensors performs very well with proper sensor deployment.
Habitat monitoring is an important driving application for wireless sensor networks (WSNs). Although researchers anticipate some challenges arising in the real-world deployments of sensor networks, a number of problems can be discovered only through experience. This paper evaluates a sensor network system described in an earlier work and presents a set of experiences from a four month long deployment on a remote island o# the coast of Maine. We present an in-depth analysis of the environmental and node health data. The close integration of WSNs with their environment provides biological data at densities previous impossible; however, we show that the sensor data is also useful for predicting system operation and network failures. Based on over one million data and health readings, we analyze the node and network design and develop network reliability profiles and failure models.