An Evaluation of the Performance of DPWS on Embedded Devices in a Body Area Network

In view of the aging society, intelligent devices pervading everyday life are faced with important challenges, such as the ease of use and the ease of configuration. The whole potential of using a body area network with several sensors to monitor vital functions of a human body can only be tapped, if the sensors used are highly specialized and tightly integrated to collaborate in a decentralized way and exhibit true plug-and-play behavior. Although the aspect of interconnecting vital sensors close to the human body bears a variety of technical challenges in itself, the development of the necessary abstraction layer in software to hide the heterogeneity of the highly specialized sensor boards is confronted with even higher challenges as these devices are often equipped with very limited resources to reduce power consumption. However, this abstraction layer is a necessary prerequisite to facilitate the development of software for a body area network with the previously mentioned characteristics. This paper presents the results of a study conducted to evaluate the performance and overhead of using web services on embedded devices to implement an abstraction layer for a body area network. In several experiments, two different implementations of the Devices Profile for Web Services (DPWS) were evaluated: the Microsoft . NET Micro Framework and the open-source DPWS-plugin from the Web Services for Devices initiative. These were used to measure absolute latencies and the "web service overhead" in the communication between three different types of resource-constraint devices.