Classification between Failed Nodes and Left Nodes in Mobile Asset Tracking Systems †

Abstract

Medical asset tracking systems track a medical device with a mobile node and determine its status as either in or out, because it can leave a monitoring area. Due to a failed node, this system may decide that a mobile asset is outside the area, even though it is within the area. In this paper, an efficient classification method is proposed to separate mobile nodes disconnected from a wireless sensor network between nodes with faults and a node that actually has left the monitoring region. The proposed scheme uses two trends extracted from the neighboring nodes of a disconnected mobile node. First is the trend in a series of the neighbor counts; the second is that of the ratios of the boundary nodes included in the neighbors. Based on such trends, the proposed method separates failed nodes from mobile nodes that are disconnected from a wireless sensor network without failures. The proposed method is evaluated using both real data generated from a medical asset tracking system and also using simulations with the network simulator (ns-2). The experimental results show that the proposed method correctly differentiates between failed nodes and nodes that are no longer in the monitoring region, including the cases that the conventional methods fail to detect.

An Efficient Recommendation Filter Model on Smart Home Big Data Analytics for Enhanced Living Environments

“…Although smart homes can provide people great convenience, they are mostly equipped with very complicated closed configurations and user interfaces [10,11]. The inhabitants living in such a smart home have to waste a great amount of time to set and operate these complicated devices and software suites [12,13].…”