PROFILE-BASED LOCATION-AWARE INFORMATION RETRIEVAL FOR MOBILE CLIENTS
Advances in wireless and mobile computing allow mobile clients to perform a wide range of applications. In certain applications, mobile clients may be interested in getting information periodically. For example, a user may need weather information every hour. This thesis addresses the problem of computing profiles (queries) efficiently and the caching of results based on the location of clients as they move. To this end, we propose a scheme that offers personalized, location-aware information access in mobile environments.
In the proposed scheme, the nodes on the fixed network execute and manage profiles on behalf of mobile clients. A simple but effective profile specification language to capture user intent has been used for retrieving information from disparate sources such as WWW and relational databases. The results for the profiles are computed and cached at appropriate nodes (based on the location of clients) and pushed to the user in a timely manner according to user specifications. The execution of the profiles is maintained on a node close to the mobile client in order to reduce the communication cost of result notification. In the proposed scheme, profiles and their executions are migrated to minimize data transmission costs. Simulation results show that the proposed scheme performs better than the existing approaches when bandwidth usage and percentage data loss are considered.
Active capability (ECA rules) has been adapted to process profiles and communicate notifications efficiently. Several mobile clients submit profiles to access location-aware information. As the execution of the profiles is done close to the location of the mobile clients, there will be considerable number of similar profiles that require processing on the same node in the fixed network. The proposed scheme groups similar queries to share common computation.