Authors:
Alexander Dannies
;
Javier Palafox-Albarrán
;
Walter Lang
and
Reiner Jedermann
Affiliation:
University of Bremen, Germany
Keyword(s):
Java, Data Processing, Wireless Sensors, OSGi, Dynamic Components, Pervasive Computing.
Related
Ontology
Subjects/Areas/Topics:
Ambient Intelligence
;
Distributed Intelligent Agents
;
Mobile and Pervasive Computing
;
Mobile Computing
;
Pervasive Embedded Networks
;
Telecommunications
;
Ubiquitous Computing Systems and Services
Abstract:
A wireless sensor network (WSN), which is one type of pervasive system, has the goal of networking heterogeneous systems and communicating through a gateway. However, it is also necessary to provide dynamic features to wireless nodes for updating applications and services during runtime. Dynamic updates can be handled either by the intrinsic features of Java or by advanced frameworks such as MIDP or OSGi. This paper investigates the software background and the feasibility of these three options in the context of WSNs. Java Virtual Machines were tested on sensor nodes and gateways currently available on the market. Two synthetic benchmarks were utilized to compare their performance. In addition, we tested the performance of an exemplary algorithm for a real life application during transportation in food logistics. Our experimental results showed that the performance of the benchmarks varied by a factor of more than 50, depending on the platform. Nevertheless, our chosen example algori
thm could be executed on all platforms within an acceptable amount of CPU time. Pre-processing of data can be applied on wireless devices to reduce communication volume and provide conclusions instead of raw data. However the use of advanced frameworks, enabling extended dynamisation, are so far very limited.
(More)