How High Should We Go? Determining Reservation Values to Negotiate Successfully for Composite Software Services

• Published in: Information systems research 2017-06, Vol.28 (2), p.353-377
• Main Authors: Sun, Sherry X., Zhao, Jing, Sarkar, Sumit
• Format: Article
• Language: English
• Subjects: automated negotiation; multiobjective optimization; Quality of Service; service composition
• ISSN: 1047-7047; 1526-5536
• DOI: 10.1287/isre.2016.0678

In the paradigm of service-centric computing, new value-added applications can be developed dynamically and flexibly by combining and integrating existing services. While software applications are traditionally specified and implemented as a set of functions uniform to all users, this new paradigm allows the same software service to be delivered with a different price, response time, availability, and other nonfunctional attributes to accommodate different modes of use. These nonfunctional attributes together are referred to as Quality of Service (QoS). When creating a new composite service, negotiation makes it possible for a service provider to offer the service with the QoS properties customized to the needs of a user. Automated negotiation tactics require the specification of reservation values for the QoS attributes. We present a methodology that determines the reservation values a user (or broker) should use for each component service based on the user’s minimum requirements for the composite service. Our methodology maximizes the chance of reaching a successful negotiation outcome while staying within the user’s reservation values for the composite service. We show that the problem of determining the user’s reservation values for component services can be modeled as a multiobjective optimization problem and then transformed to a single-objective optimization problem using a max-min approach. The formulation can incorporate providers' different QoS preferences to increase the chance of negotiation success. We identify problem instances for which closed-form solutions can be found for the reservation values. We show how the method of setting reservation values can be incorporated into a negotiation process that uses extant concession and trade-off tactics. Simulation experiments demonstrate the effectiveness of the proposed approach. If some providers accept offers before the negotiation process deadline, we show that dynamically changing the reservation values for the remaining providers makes the overall negotiation process more likely to succeed. The online appendix is available at https://doi.org/10.1287/isre.2016.0678 .