W-ADE: Timing Performance Benchmarking in Web of Things

Abstract

As the number of devices participating in the Internet of Things (IoT) rapidly grows, the challenge of interoperability across IoT platforms becomes more apparent. In order to limit fragmentation of IoT development and improve compatibility, web mechanisms and technologies can be applied, forming the Web of Things (WoT). The World Wide Web Consortium (W3C) supports the standardization of WoT by providing a platform-independent specification called Thing Description (TD). It is a machine-readable document that semantically describes metadata, interactions and interfaces of a device, indicating its functionality. However, it does not provide any information about timing performance, which is crucial for the design of optimal system compositions. In this paper, we present W-ADE, a development environment for WoT and TD that facilitates manual timing measurements and automated timing performance benchmarking of Thing interactions, merely with a TD available. Timing performance is guaranteed systematically, hence allowing optimization during the design phase of Thing mashups. Our evaluation shows that with 99.9% confidence W-ADE can predict average interaction timing performance within a range of \({\pm }5\%\), and is able to provide approximate network-independent static timing performance benchmarks for interaction affordances to 99.93%. To enable the design of heterogeneous IoT applications based upon these timing requirements, a proposal on how to annotate a TD based on the measured performance data is made.