Research On Agent Based Service Trading Model And Negotiation Algorithm

Stepping into the 21st century, the development models of manufacturing enterprises are encountering tremendous changes. The specialized and servitizational division of labor and collaboration between enterprises demand them to continuously acquire and focus their kernel resources to their specialized areas and thus they can establish their distinctive positions and values among others. At the same time, enterprise users also need to utilize capabilities provided by other enterprises in an on-demand way, according to the rules of demand-driven, user being dominant, on-demand service, pay-per-use and dissemination after use. In this way, enterprises can achieve the integrated application and coordination of resources dynamically and broadly.The rapid growth of new technologies such as Cloud computing and Internet of Things provide a new way to both explore service modes in cloud manufacturing and solve the various problems which are encountered in the development of manufacturing industry in our country. Under the cloud manufacturing mode, enterprises provide and use active, agile and aggregated manufacturing resources and capabilities based on market mechanisms. Service providers and consumers make mutual choices to engage in a service trading. Service providers need to decide whether to engage into a service trading and which kind of service to provide to make how much profit, while service consumers can select a service provider who can meet his demand best to trade a service. In such market-oriented environments, it has become a problem that how to establish an effective model and mechanism to support enterprises to act as providers and consumers to trade services.Since service computing is a new research area, no effective approach for solving service trading problems has been proposed. Due to the situations that the character of a service is often decided by multiple attributes, and grids and clouds operate under open and dynamic environments, the problem of service trading cannot be easily solved. And traditional market mechanisms for e-commerce and resource trading in distributed computing, such as fixed-pricing and auctions, cannot meet the requirements of multi-attribute based service trading.With the research background of service trading between manufacturing enterprises, focusing on the key technologies of service trading in grid and cloud computing, under the support of The National Natural Science Foundation of China (NSFC) "Research on Automobile Collaboration Manufacturing Grid and Applications", National High Technology Research and Development Program ("863"Program) "Service Platforms for Supporting Business Collaboration in Machinery Industry Clusters", research program "Research on Key Technologies of Cloud Platform for Small and Medium Sized Enterprises" and some other research programs, this thesis proposes models and techniques for service trading and multi-attribute service negotiation problems in manufacturing grids and cloud manufacturing. The main contributions of the thesis are as follows:1. An Agent based Service Trading ModelWe present a service trading model based on agent technologies, in which service providers and consumers delegate their trading tasks to agents. By doing this, the service provider agents and consumer agents can conduct automated service negotiation on behalf of their human users, which can improve trade efficiency by reducing time and cost, and avoid errors induced by human operations. During the negotiation process, the bargaining issues include not only price, but also other service attributes such as service qualities, delivery time, service reliability and etc. In this model, the trading object, i.e., the service is expressed in Ontology Web Language. Through negotiation, service providers and consumers establish a contract to govern the service delivery to complete a service trading process.2. Pareto-front Negotiation Approach Based on NSGA-II Algorithm and Fair Win-Win Negotiation ApproachIn our negotiation model, the negotiation process between service providers and consumers is divided into two phases, which are Pareto-front negotiation in the first phase and bilateral bargaining in the second phase. We propose an improved and distributed NSGA-II algorithm to support service providers and consumers to find Pareto-fronts through bilateral negotiation, which does not require them to disclose their private utility functions, nor do they need to rely on a third party mediator. Besides, we also propose an approach for service providers and consumers to achieve fair win-win agreements based on the Pareto-fronts obtained in the first phase through a one-step negotiation.3. Simultaneous-Offer Bilateral Negotiation Model and Concession Strategy ModelTo eliminate the deficiencies of current alternating-offer bilateral negotiation protocol, we propose a simultaneous-offer bilateral negotiation model based on a third party mediator. Also, we propose a new time-dependent concession strategy model, which solves the problem that agents do not consider its opponent’s offers in making a concession decision in current existing time-dependent concession models.4. A One-to-Many Concurrent Service Negotiation modelWe propose a one-to-many concurrent negotiation model, which can support a service consumer to negotiate with multiple service providers concurrently. This concurrent negotiation model improves the efficiency for the multiple negotiations, saves time and also optimizes the final negotiation results.Through theoretical and experimental analysis, the agent based service trading model, the multi-attribute simultaneous-offer bilateral negotiation model and concurrent negotiation model proposed in this thesis meet the requirements of service negotiation and trading problem in manufacturing girds and cloud manufacturing. Through comparison experiments, the negotiation model proposed in this thesis are proved to work better than current existing models. Based on the above work, some research results are already applied to our "Machinery Industry Service Platform". Other research and techniques, such as coordination of multiple negotiations over multiple interdependent services, adaptive and automated adoption of negotiation strategies in reacting to market environments, and so on are the future work.