A Cooperative Game for Multi-Agent Collaborative Planning

Cooperative game is an approach for a group of decision-making agents (DMAs) to reach mutually beneficial agreements where the players make agreements in order to optimize their common cost or payment. A coordination mechanism is required to achieve a rational and structured plan for a decision jointly made by independent DMAs. This paper presents a cooperative game for the problem of collaborative planning where each planning domain is controlled by a DMA. The main purpose here is to demonstrate how the cooperative game theory can be applied to the problem of multi-agent collaborative planning. It deals with cost allocation methods which is basically an optimization problem. We present several coordination mechanisms for the game based on domain planning, data exchange and compensation negotiation. These belong to two categories, namely, multistage local planning domain based coordination mechanism (LPDCM) and single stage global planning domain based coordination mechanism (GPDCM). Here DMAs wants to maintain the privacy of their strategic information. This asymmetry of information may cause increase in cost and generate inefficient solutions. The mechanisms preserve the privacy of information using secure multi-party computation concepts. This improves the quality of a plan and subsequent decision-making process of the cooperative game significantly. This work introduces a secure multi-party linear programming extended protocol (SMLPEP) by extending an existing two-party secure linear programming protocol. This work extends the Chakraborty’s work on collaborative planning for supply chain [17] and [3], which in turn was based on Dudek’s piece [6] of 2004. The present work looks at the collaborative planning problem of [6] from the perspective of cooperative game, focuses on the development of mechanism design for efficient cost sharing purposes for several scenarios and thus enhances the generalization capability of the