Online Mining in Sensor Networks

Abstract

Abstract. Online mining in large sensor networks just starts to attract interest. Finding patterns in such an environment is both compelling and challenging. The goal of this position paper is to understand the challenges and to identify the research problems in online mining for sensor networks. As an initial step, we identify the following three problems to work on: (1) sensor data irregularities detection; (2) sensor data clustering; and (3) sensory attribute correlations discovery. We also outline our preliminary proposal of solutions to these problems.

Outlier Detection Techniques for Wireless Sensor Networks: A Survey

“…In other words the main question is how to process as much data as possible in a decentralized and online fashion while keeping the communication overhead, memory and computational cost low (Ma et al, 2004).…”

“…A wide variety of applications of WSNs includes those relating to personal, industrial, business, and military domains, such as environmental and habitat monitoring, object and inventory tracking, health and medical monitoring, battlefield observation, industrial safety and control, to name but a few. In many of these applications, real-time data mining of sensor data to promptly make intelligent decisions is essential (Ma et al, 2004).…”

A Communication-Efficient Distributed Clustering Algorithm for Sensor Networks

“…There have been few attempts to address the problem of clustering sensory data. Notable exceptions are the ones proposed in [4,12] for clustering in a multi-dimensional sensor dataset. But these approaches do not consider the physical limitations of the sensor networks.…”

“…Moreover, data which is sensed in WSNs is generated and streamed continuously. These challenges make traditional mining techniques inapplicable in sensor networks [4], as the traditional mining process is usually centralized, computationally expensive, and focuses on disk stored data.…”

SO_MAD: SensOr Mining for Anomaly Detection in Railway Data

“…Different approaches focusing either on the data representation (e.g., sensors clustering [1], discretization [2]) or knowledge extraction (e.g., association rules [2], [3], [4], [5], sequential patterns [6], [7], [8]) were proposed. Nevertheless, they usually do not consider that contextual information could improve the quality of the extracted knowledge.…”