An Architecture to Support Learning-based Adaptation of Persistent Queries in Mobile Environments

Abstract

Queries are frequently used by applications in dynamically formed mobile networks to discover and acquire information and services available in the surrounding environment. A number of inquiry strategies exist, each of which embodies an approach to disseminating a query and collecting results. The choice of inquiry strategy has different tradeoffs under different operating conditions. Therefore, it is beneficial to allow a query-based application to dynamically adapt its inquiry strategy to the changing environmental conditions. To promote development by non-expert domain programmers, we can automate the decision-making process associated with adapting the inquiry strategy. In this paper, we propose an architecture to support automated adaptative query processing for dynamic mobile environments. The decision-support module of our architecture relies on an instance-based learning approach to support context-aware adaptation of the inquiry strategy.