Please use this identifier to cite or link to this item: http://hdl.handle.net/10397/75251
Title: Quality assurances of probabilistic queries
Authors: Cheng, R 
Chan, E
Lam, KY
Issue Date: 2007
Publisher: Springer
Source: In NP Mahalik (Ed.), Sensor networks and configuration : fundamentals, standards, platforms, and applications, p. 189-210. Berlin: Springer, 2007 How to cite?
Abstract: Many applications use sensors extensively to capture and monitor the status of physical entities. In a habitat-monitoring system, for example, the temperature values of birds nests are investigated through the use of wireless sensor networks (Deshpande 2004). Sensors are also installed in different parts of the building, so that the temperature offices can be adjusted by an internal air-conditioning system. In fact, sensors are increasingly used in various applications, which deliver monitoring and querying services based on various attribute values of physical environments, such as location data, temperature, pressure, rainfall, wind speed, and UV-index. In this kind of systems, one common problem is that the reading of a sensor can be uncertain, noisy and error-prone (Elnahrawy 2003, Cheng 2003). A sensors value can be contaminated by measurement errors. Moreover, the environment being monitored by sensors can change continuously with time, but due to limited battery power and network bandwidth, the state of the environment is only sampled periodically. As a result, the data received from the sensor can be uncertain and stale. If the system uses these data, it may yield incorrect information to users and make false decisions. In order to solve this problem, the uncertainty of the sensor data must be taken into account in order to process a query (Cheng 2003). In order to consider uncertain data during query execution, the concept of probabilistic queries has been studied extensively in recent years (Wolfson 1999; Pfoser 2001; Cheng 2003; Desphande 2004). In these works, uncertain data are modeled as a range of possible values with a certain probability distribution (e.g., Gaussian and uniform distribution). Probabilistic queries process these uncertain data and produce "imprecise answers". These answers are those that are augmented with probabilistic guarantees to indicate the confidence of answers. As an example, consider a query asking "which area yields a wind speed over 1 km/hr", where the wind speed values of some regions are reported by sensors. By modeling each wind speed value as an uncertain data item, a probabilistic query returns a list of areas, together with their probabilities, which indicate the chance that the area yields a wind speed higher than 1 km/hr. Notice that although probabilistic queries do not return exact answers, the probability values reflect the degree of confidence for the answer, rather than a completely wrong answer when uncertainty is not considered. In fact, the probability values augmented to a probabilistic query answer can serve as some indicators for the quality of query answers. Consider a MAX query, which returns all objects which have non-zero probability of giving a maximum value. This query is executed over two different sets of data, namely {A, B, C} and {D, E, F}, and yields the following answers.
URI: http://hdl.handle.net/10397/75251
ISBN: 9783540373667 (electronic bk.)
9783540373643 (paper)
DOI: 10.1007/3-540-37366-7_9
Appears in Collections:Book Chapter

Access
View full-text via PolyU eLinks SFX Query
Show full item record

Page view(s)

12
Last Week
3
Last month
Citations as of Nov 11, 2018

Google ScholarTM

Check

Altmetric


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.