Figure 6 - uploaded by Nalini Venkatasubramanian
Content may be subject to copyright.
Source publication
We consider the problem of evaluating continuous selection queries over sensor-generated values in the presence of faults. Small sensors are fragile, have fl- nite energy and memory, and communicate over a lossy medium; hence, tuples produced by them may not reach the querying node, resulting in an incomplete and ambiguous answer, as any of the non...
Similar publications
In IP-based media delivery services, we often deal with predictable network load and traffic, making it beneficial to use advance reservations even when network failure occurs. In such a network, to offer reliable reservations, fault-tolerance related features should be incorporated in the advance reservation system. In this paper, we propose an op...
Deriving fault-tolerant schedulability resilience for real-time systems has been a challenge. While some approaches are based on worst-case condi-tions, some others are designed to a specific scheduling policy. A simulation-based analysis has recently been proposed as an attempt to overcome such short-comings. Its basic idea is to simulate the syst...
Fault tolerance, reliability and resilience in Cloud Computing are of paramount importance to ensure continuous operation and correct results, even in the presence of a given maximum amount of faulty components. Most existing research and implementations focus on architecture-specific solutions to introduce fault tolerance. This implies that users...
Citations
... In addition to supporting these reliability specifications, we have developed more informative reliability metrics with well-defined semantics. For instance , the recall metric used in information retrieval may be used to indicate the desired completeness of the answer set, if the application is gathering all the readings that meet certain conditions [Lazaridis et al. (2006)]; reliability can also be specified as tolerable thresholds on " false-alarm " or " missed-event " probabilities (i.e., bounds on detection or estimation accu- racy) [Hwang et al. (2005)]. @BULLET QoD (Quality of Data) desired from the sensing substrate may be imposed on individual sensor values, or on an answer computed over readings from a set of sensor reports. ...
Many sensor applications often require collecting raw sensed values from many sensor nodes to one centralised server. Sensor data collection typically comes with various quality requirements, e.g. the level of precision requested for temperature values, the time constraints for getting the data, or the percentage of data that is needed. This paper presents a quality-aware sensing framework where characterisations of sensor applications' quality needs are identified and different sensor data collection problems are classified. Two problems and their solutions are then presented as examples to demonstrate how single (or multiple) quality need(s) are satisfied. The paper concludes with suggestions for future research directions that have the potential to complete the framework and provide a holistic approach to sensor applications with diverse quality requirements.
In this chapter, we provide a data management perspective on large-scale sensor environments applications posing non-functional requirements to meet the underlying timeliness, reliability and accuracy needs in addition to the functional needs of data collection. Due to the large-scale regional spread, we need methods that will allow scaling of today’s systems to large-scale deployments. Our data management techniques have solved a fundamental challenge in such situations, that is the ability to handle the explosion of sensor data in sensor networks, either due to scaling of the network or due to increased data generation by highly capable and “media-rich” nodes.
