Composition of Web Services Using Markov Decision Processes and Dynamic Programming

الملخص EN

We propose a Markov decision process model for solving the Web service composition (WSC)problem.

Iterative policy evaluation, value iteration, and policy iteration algorithms are used toexperimentally validate our approach, with artificial and real data.

The experimental resultsshow the reliability of the model and the methods employed, with policy iteration being the bestone in terms of the minimum number of iterations needed to estimate an optimal policy, with thehighest Quality of Service attributes.

Our experimental work shows how the solution of a WSCproblem involving a set of 100,000 individual Web services and where a valid compositionrequiring the selection of 1,000 services from the available set can be computed in the worstcase in less than 200 seconds, using an Intel Core i5 computer with 6 GB RAM.

Moreover, a realWSC problem involving only 7 individual Web services requires less than 0.08 seconds, using thesame computational power.

Finally, a comparison with two popular reinforcement learningalgorithms, sarsa and Q-learning, shows that these algorithms require one or two orders ofmagnitude and more time than policy iteration, iterative policy evaluation, and value iteration tohandle WSC problems of the same complexity.