Data Quality and Energy Management Tradeoffs in Sensor Service Clouds

Cloud-based sensor data collection services are becoming an essential part of the Internet of Things (IoT). As the consumer demand grows for these services, the data quality (DQ) of the stream becomes an increasingly vital issue. Of particular interest is the inherent tradeoff between the DQ and the energy consumption of the sensor. Unfortunately, there has been very little research on the management of this tradeoff that allows data consumers to receive high quality data while simultaneously conserving energy. Our work seeks to explore this tradeoff in detail by combining DQ services for the data stream consumer with customizable energy efficient "EE" throttling algorithms for the data feed producers. These energy management services provide cost reduction rewards for consumers who would otherwise make poor DQ/EE decisions. Our primary contributions include cloud-based services for monitoring the tradeoff, an architecture that adjusts to DQ needs and a producer/consumer data stream best matching cloud service. We envision that our services architecture will reward energy efficiency decisions and profoundly affect consumer choices.

I. Introduction

Big data issues in the IoT are characterized by continuous streams of data being generated by millions of geographically distributed sensor devices [1]. Most of the current research on big data has focused on scalability, storage and data analytics. Architectures like Hadoop and Apache Storm as well as large scale storage with NoSQL databases like MongoDB are addressing these issues. However, an equally important concern that has hitherto received very little research attention is that of data quality. In fact high quality data is fundamental to deriving sound and robust business intelligence in many domains. Our research contends that DQ is becoming one of the primary contributors to advancement of Big Data research.