| The use of Service Oriented Architectures, namely web services, has become a widely adopted method for transfer of data between systems across the Internet as well as the Enterprise. Adopting a similar approach to embedded devices is also starting to emerge as personal devices and sensor networks are becoming more common in the industry. This research examines web service encoding mechanisms currently in place and their system resource utilization on both a full computer system and an embedded device. Research was conducted to first baseline SOAP and JSON serialization and deserialization performance, and then develop a web service architecture for an embedded device and verify similar performance characteristics exist. In the process of developing the second experiment, a flexible Python-based web service architecture is developed and implemented on an embedded device. Performance of SOAP, XML over HTTP, and JSON encoding mechanisms are measured and compared in this environment. Attempts to implement a similarly functioning architecture using a Java framework called XINS failed due to resource constraints imposed by the embedded device.;In every case tested, JSON encoding significantly outperformed the tested XML-based encoding, showing clear advantages in both processing performance and network bandwidth utilization. Also shown is that XML over HTTP has performance characteristics similar to that of JSON for smaller, less complicated data objects, but JSON is significantly more efficient when more complicated data objects are used. |
