Distributed weighted clustering of evolving sensor data streams with noise

摘要: Collecting data from sensor nodes is the ultimate goal of Wireless Sensor Networks. This performed by transmitting sensed measurements to some collecting station. In nodes, radio communication dominating consumer energy resources which are usually limited. Summarizing internally on and sending only summaries will considerably save energy. Clustering an established mining technique for grouping objects based similarity. For networks, k-center clustering aims at in groups, each contains similar measurements. this paper we propose a novel resource-aware -center algorithm called: SenClu. Our immediately detects new trends drifting stream follows them. SenClu powerfully uses light- weighted decaying that gives lower influence old data. As noisy, our also outlier-aware. thorough experiments synthetic real world sets, show outperforms two state-of-the-art algorithms producing higher quality following stream, while consuming nearly same amount