An adaptive context-aware optimization framework for multimedia adaptation service selection

In pervasive systems, context is a direct cause to adapt the content of multimedia documents so that they comply, as far as possible, with the current constraints. In this respect, several adaptation approaches have already been proposed, in which adaptation services are often selected from shortlists of services. Practically speaking, adaptation services are provided in various instances and ways, thus making the selection task more difficult. Furthermore, existing approaches for the service selection paradigm cannot be properly applied mainly because constraints on execution time and the availability of computation resources must be considered. To deal with this issue, we propose a framework for adaptive service selection using a bag of metaheuristics ranging from local to global search methods. Depending on the contextual constraints, a sub-bag of algorithms is selected, for which the budget is distributed, using a reinforcement learning mechanism related to their performances. The proposal is validated through a set of experiments and comparisons. The obtained results are satisfactory and encouraging.