Efficient Nonlinear Regression-Based Compression of Big Sensing Data on Cloud
- Publication Type:
- Chapter
- Citation:
- Big Data Analytics for Sensor-Network Collected Intelligence, 2017, pp. 83 - 98
- Issue Date:
- 2017-02-08
Closed Access
| Filename | Description | Size | |||
|---|---|---|---|---|---|
| chapter.pdf | Published version | 5.74 MB |
Copyright Clearance Process
- Recently Added
- In Progress
- Closed Access
This item is closed access and not available.
© 2017 All rights reserved. With the advance of modern technology, big sensing data is commonly encountered in every aspect of industrial activity, scientific research and people's lives. In order to process that big sensing data with the computational power of the cloud effectively, different compression strategies have been proposed including data trend-based approaches and linear regression-based approaches. However, in lots of real-world applications, the incoming big sensing data can be extremely bumpy and discrete. Thus, at the big data processing steps of data collection and data preparation, the above compression techniques may lose effect in terms of scalability and compression due to the inner constraints of their predicting models. To improve the effectiveness and efficiency for processing those real-world big sensing data, in this chapter, a novel nonlinear regression prediction model is introduced. The related details, including regression design, least squares, and triangular transform, are also discussed. To explore fully the power and resource offered by the cloud, the proposed nonlinear compression is implemented with MapReduce for achieving scalability. Through our experiment based on real-world earthquake big sensing data, we demonstrate that the compression based on the proposed nonlinear regression model can achieve significant storage and time performance gains compared to previous compression models when processing similar big sensing data on the cloud.
Please use this identifier to cite or link to this item: