Mid Sweden University

As the complexity of data sources increases, it becomes a significant challenge to develop and maintain applications which are required to interact with heterogeneous data sources. It becomes even more difficult when the intention of the application is to frequently modify the data source. Application developers must look into thousands lines of source code to locate the corresponding part and to modify them one by one. The diversity of data source properties and data schemas makes it complicated for the application to adapt to a new data source.

In this thesis, a middleware solution is proposed to address this problem. Both permanent and real-time data should pass through the middleware during communication between the data source and the application, or among data sources.

The middleware creates a data source service wrapper for each data source and uses an abstract class to shield different wrappers. A global-local mapping mechanism is used to eliminate the data schema incompatibility. In addition, xml technology is used to accomplish data transmission, which eases the interface design and ensures platform independency. At the same time, a Data Source Wizard has been developed to assist the maintainer of the middleware with items for middleware extensions. The Data Source Wizard makes it more convenient and standardized to add a new data source.

The experience of integrating the middleware into this Android E-Health application has proved the flexibility and extensibility of the middleware. Because the middleware does not contain any specific business logic, it can be generally applied to other applications with different purposes.