Affiliations: [a] School of Computer Engineering, Hubei University of Arts and Science, Hubei, China | [b] Computer School, Yangtze University, Hubei, China
Correspondence:
[*]
Corresponding author: Min Wang, School of Computer Engineering, Hubei University of Arts and Science, Hubei 441053, China. E-mail: [email protected].
Abstract: Poor quality and harsh condition can lead to the existence of error and abnormal data in sampling data of sensor nodes. Therefore, sometimes the average value query (AVG query) does not reflect the real average level of the monitoring area. In this case, median query is needed to reflect the average level of the monitoring area. In this paper, we first propose a median query algorithm HMA based on equal height histogram. We can establish the global histogram to determine which interval the median falls, and then we only collect data in the interval, so as to reduce the data we need to transmit. Then we extend HMA algorithm and put forward HFMA algorithm combining histogram and filter. In HFMA we only need to collect the data inside filter and aggregate influence coefficient during each sampling period. The base station can calculate the median according to the collected data and the aggregated value of the influence coefficient. Experimental results have shown that the HFMA algorithm is superior to the NAIVE algorithm and the HMA algorithm, which can effectively save the energy cost and improve the network lifetime.
Keywords: Sensor network, continuous median query, filter, influence coefficient
