Price controlled resource allocation for the provision of information products and services employing combinatorial auctions

Abstract The dynamic allocation of resources for the supply of Information Services and Information Products (ISIP) is of increasing importance for infrastructure and service providers in growing B2B and B2C markets. Based on the FIPA oriented Multi Agent System (MAS) platform JADE we developed a model which simulates the allocation of ISIP resources by a Combinatorial Auction (CA). An Improved Greedy CA-Algorithm (IG-CAA) and a Simulated Annealing based CA-Algorithm (SA-CAA) are proposed to solve the resulting winner determination problem. Using the bid price for task processing as a control variable has turned out to be an efficient tool even in non-economic settings. Performing multiple simulation runs, we could show the superiority of the SA-CAA to the IG-SAA for a test scenario consisting of unstructured bids. However we failed to demonstrate, that the SA-CAA is capable to handle a satisfying ISIP resource allocation in distributed systems with affordable computational expense and could only recommend the IG-CAA for this purpose, due to its lower calculation requirements.