TMS-RFID: Temporal management of large-scale RFID applications |
| |
Authors: | Xue Li Jing Liu Quan Z Sheng Sherali Zeadally Weicai Zhong |
| |
Affiliation: | (1) School of Information Technology and Electrical Engineering, The University of Queensland, Brisbane, Australia;(2) Institute of Intelligent Information Processing, Xidian University, Xi’an, China;(3) School of Computer Science, The University of Adelaide, Adelaide, Australia;(4) Department of Computer Science and Information Technology, University of the District of Columbia, Washington, D.C., USA |
| |
Abstract: | In coming years, there will be billions of RFID tags living in the world tagging almost everything for tracking and identification
purposes. This phenomenon will impose a new challenge not only to the network capacity but also to the scalability of event
processing of RFID applications. Since most RFID applications are time sensitive, we propose a notion of Time To Live (TTL), representing the period of time that an RFID event can legally live in an RFID data management system, to manage various
temporal event patterns. TTL is critical in the “Internet of Things” for handling a tremendous amount of partial event-tracking
results. Also, TTL can be used to provide prompt responses to time-critical events so that the RFID data streams can be handled
timely. We divide TTL into four categories according to the general event-handling patterns. Moreover, to extract event sequence
from an unordered event stream correctly and handle TTL constrained event sequence effectively, we design a new data structure,
namely Double Level Sequence Instance List (DLSIList), to record intermediate stages of event sequences. On the basis of this,
an RFID data management system, namely Temporal Management System over RFID data streams (TMS-RFID), has been developed. This
system can be constructed as a stand-alone middleware component to manage temporal event patterns. We demonstrate the effectiveness
of TMS-RFID on extracting complex temporal event patterns through a detailed performance study using a range of high-speed
data streams and various queries. The results show that TMS-RFID has a very high throughput, namely 170,000–870,000 events
per second for different highly complex continuous queries. Moreover, the experiments also show that the main structure to
record the intermediate stages in TMS-RFID does not increase exponentially with the number of events. These results demonstrate
that TMS-RFID not only supports high processing speeds, but is also highly scalable. |
| |
Keywords: | |
本文献已被 SpringerLink 等数据库收录! |
|