| Summary: | Quality of Service (QoS)-aware Web Service Composition as complex problem solver has become one of the most highlighted issues in service computing area. It maps to multi-objective optimization problem that is classified as Nondeterministic Polynomial-time hard (NP-hard) problem. The diversity of subjective and potentially dishonest evaluations impose an obstacle to QoS-aware service assessment. The vague preferences of users have also to be considered in multi-criteria service selection. Last but not least, the budget-constrained service negotiation needs to make trade-off between the desired QoS metrics and the imposed budget constraints by service users. There is a large body of research covering aforementioned different aspects of service composition. This research tries to open a new horizon for service composition to utilize collaborative decision support systems. The proposed system involves three phases, namely Trust-Aware Crowd-enabled consensus-based Service Assessment (TACSA), Fuzzy inference-based multi-criteria Service Ranking (FASER), and Pareto-optimal service composition (PALEN). In the first phase, TACSA is responsible to assess all candidate services with respect to the required QoS metrics and guarantee this assessment not to suffer from subjective and dishonest evaluations by means of the collaborative decision making. The incurred complexity in capturing users' preferences and objectives is the second obstacle to rank services. FASER, the fuzzy inference engine, is then used to capture user preferences and support multi-criteria QoS-based service ranking. After that, the composer is required to negotiate with ranked service providers and select the best-possible candidate service based on users' QoS desires and constraints. PALEN enables composer to achieve this aim using the autonomous service level agreement negotiation strategy and surplus management. The focus of the proposed negotiation strategy is restricted to the time-dependent tactic that can handle the deadline imposed by users. Besides, a novel approach proposed to dynamically adjust time-dependent function parameter based on service demand and utilization, and redistribute surplus to optimize the composite service. The research promises to select the best candidate services that maximizes QoS metrics while adheres to users' budget constraints. The extensive experimental results along with simulated scenarios demonstrate the applicability and effectiveness of the proposed approach. It is interesting to note that the consensus on assessed QoS metrics is achieved with respect to different parameters and the crowd converge to the most trustworthy service assessment. Moreover, the results indicate that the composition optimality is averagely increased by almost 80% considering different composition scenarios.
|
|---|
