FDMOABC: Fuzzy Discrete Multi-Objective Artificial Bee Colony approach for solving the non-deterministic QoS-driven web service composition problem

Abstract

The multi-objective quality of service (QoS)-driven web service composition problem (MOQWSCP) aims to find the best combinations of atomic web services (i.e. composite service) to answer high quality of the optimized QoS criteria in a way that maximize benefit QoS parameters such as availability and reliability and minimize the negative ones like price and response time, where the users’ requirements should be satisfied. Due to the dynamic environments in which the elementary services are invoked, some services’ QoS parameters are often ambiguous and uncertain, so, it is inappropriate to express them by fixed values. Hence, The QoS parameters are represented by trapezoidal fuzzy numbers. Thus, we formulate MOQWSCP as a fuzzy multi-objective optimization problem (FMOQWSCP). A fuzzy discrete multi-objective artificial bee colony (FDMOABC) approach is provided to solve the formulated FMOQWSCP, for which we have integrated a new fuzzy ranking method to cope with solutions sorting and a new fuzzy distance measure that is used to control and keep the diversity of FDMOABC’s solutions. Furthermore, a fuzzy multi-criteria decision-making method (FMCDMM) is provided to determine the best composite service among the Pareto-optimal solutions generated by FDMOABC. Finally, two kinds of comparisons are performed to validate the performance and the effectiveness of FDMOABC and FMCDMM methods. In the former, the combined FDMOABC and FMCDMM methods is compared against the fuzzy single objective optimization approaches TGA and EFPA, whereas in the later, a multi-objective optimization comparison is performed among FDMOABC and the fuzzy-extended versions of NSGA-II and SPEA2 algorithms.