DDEUDSC: A Dynamic Distance Estimation using Uncertain Data Stream Clustering in mobile wireless sensor networks

• Published in: Measurement : journal of the International Measurement Confederation 2014-09, Vol.55, p.423-433
• Main Authors: Luo, Qinghua, Yan, Xiaozhen, Li, Junbao, Peng, Yu
• Format: Article
• Language: English
• Subjects: Clustering algorithm; Dynamic distance estimation; Interval data; RSSI; Uncertain data stream
• ISSN: 0263-2241; 1873-412X
• DOI: 10.1016/j.measurement.2014.05.040

•There are dynamics and uncertainty in RSSI data stream in different environments.•We express the uncertainty of RSSI values in terms of interval data.•Distribution-based mapping and clustering is used to overcome uncertainty.•Correlation relationship between cluster centers are considered in data pattern.•Data pattern is used to improve accuracy of dynamic distance estimation. In RSSI (Received Signal Strength Indicator)-based communication distance estimation of mobile wireless sensor network localization, RSSI is assumed to exponential attenuation with increment of communication distance in ideal radio propagation models, which is invalid due to the uncertainty of RSSI data in real communication environment, resulting in considerable error of communication distance estimation. Moreover, dynamic distance estimation demands a high efficiency of computation for the continual generation of RSSI data stream in the mobile node. This paper develops a new dynamic communication distance estimation method using uncertain interval data stream clustering, named as DDEUDSC (Dynamic Distance Estimation method using Uncertain Data Stream Clustering). First, statistical information of RSSI data is used to represent the RSSI-D mapping relationship in terms of interval data. Then we consider the data pattern composed of some consecutive cluster centers, and apply it in our uncertain RSSI data stream clustering algorithm to estimate the dynamic communication distance. Finally, RSSI data streams in three typical communication environments are conducted for experiments. The experimental results show the proposed method is an effective way to improve RSSI-D estimation precision in RSSI data stream with uncertainty and dynamics characteristic.