Comparison of Component-based Adaptation to In-application Adaptation

The need for adaptation in mobile and wireless environments is well established. Puppeteer is a component-based adaptation system that changes the behavior of applications for bandwidth limited networks. Puppeteer uses the APIs exported by the application and the document structure exposed by the application, and is thus limited by the overheads of APIs and the limitations of the document structure. In order to quantify the overheads of using the API, we changed the source code of applications to perform the same adaptive actions from inside the application. We refer to this approach as the In-Application adaptation system. This thesis work describes the design and implementation of an InApplication adaptation system based on SUN Microsystems OpenOffice suite. Using Puppeteer and In-Application adaptation, we perform a head-to-head comparison of the overhead one incurs using APIs, and answer some questions such as: are we limited by the APIs and document structure of applications? Performance results indicate that the In-Application adaptation system is capable of reducing latency by about 75% for application-specific adaptation policies. Comparisons reveal, however, that Puppeteer incurs only 6% more adaptation costs in the worst case as compared to the In-Application adaptation approach. The APIs exported by the application are sufficient to implement powerful adaptation policies and the net savings achieved due to latency reduction of InApplication approach are acceptable. Therefore, we do not advocate modifying complex application source code for the purposes of adaptation.