Database Reference
In-Depth Information
acquiring sensor data. We, then, discuss the application of models in
sensor data cleaning; followed by a discussion on model-based meth-
ods for querying sensor data. Lastly, we survey model-based methods
proposed for data compression and synopsis generation.
Keywords: model-based techniques, data acquisition, query processing, data clean-
ing, data compression.
1. Introduction
In recent years, there has been tremendous growth in the data gen-
erated by sensor networks. Equivalently, there are pertinent techniques
proposed in recent literature for eciently acquiring and managing sen-
sor data. One important category of techniques that have received sig-
nificant attention are the model-based techniques. These techniques use
mathematical models for solving various problems pertaining to sensor
data acquisition and management. In this chapter, we survey a large
number of state-of-the-art model-based techniques for sensor data ac-
quisition and management. Model-based techniques use various types of
models: statistical, signal processing, regression-based, machine learn-
ing, probabilistic, or time series. These models serve various purposes
in sensor data acquisition and management.
It is well-known that many physical attributes, like, ambient tempera-
ture or relative humidity, vary smoothly. As a result of this smoothness,
sensor data typically exhibits the following properties: (a) it is continu-
ous (although we only have a finite number of samples), (b) it has finite
energy or it is band-limited, (c) it exhibits Markovian behavior or the
value at a time instant depends only on the value at a previous time
instant. Most model-based techniques exploit these properties for e-
ciently performing various tasks related to sensor data acquisition and
management.
In this chapter, we consider four broad categories of sensor data man-
agement tasks: data acquisition, data cleaning, query processing, and
data compression. These tasks are pictorially summarized in the toy
example shown in Figure 2.1 . From Figure 2.1 , it is interesting to note
how a single type of model (linear) can be used for performing these
various tasks. For each task considered in this chapter, we extensively
discuss various, well-researched model-based solutions. Following is the
detailed discussion on the sensor data management tasks covered in this
chapter:
Search WWH ::




Custom Search