Simultaneous service selection for multiple composite service requests: A combinatorial auction approach

Abstract

Emerging developments in service-based systems have contributed to a range of proposals for the rapid development of complex applications by composing existing web services through standards-based interfaces. With the availability of a large number of such services offering identical functionality at varying levels of price and other Quality-of-Service (QoS) parameters, selecting the best web services for composition is critical– the problem referred to as composite service selection (CSS). Extant CSS approaches have tended to offer solutions to the problem for a single composite service. We extend CSS to address the setting involving multiple, simultaneous composite service requests. We propose a service selection mechanism based on combinatorial auctions to solve this problem by simultaneously matching the web service offers and requests and performing extensive evaluations in four market sectors with varying economic sizes and complexity of requests. The results indicate that the proposed simultaneous mechanism achieves a higher success rate in allocating services to requesters compared to sequential matching mechanisms. While the average cost difference in procuring composite services under the two mechanisms is not statistically significant, the simultaneous mechanism achieves more stable and homogenous costs over time.