An integrated simulation, Markov decision processes and game theoretic framework for analysis of supply chain competitions
Dong Xu, Young‐Jun Son · Winter Simulation Conference · 2013
The proposed framework is composed of 1) simulation-based game platform, 2) game solving and analysis module, and 3) Markov decision processes (MDP) module, which are illustrated for a supply chain system under the newsvendor setting through two different phases. At phase 1, the simulation-based game platform is firstly constructed to formulate both the supply chain horizontal and vertical competitions. A novel game solving and analysis procedure is proposed to include 1) strategy refinement, 2) data sampling, 3) gaming solving, and 4) solution quality evaluation. At phase 2, the problem is extended into multi-period setting, in which discrete-time MDP with the discounted criterion is employed. The influences of each agent's competitor decision are incorporated so a stochastic game is formulated, in which a multi-agent reinforcement learning technique is applied as a solution approach. Experimental results demonstrate the system performance, MDP solving effectiveness and efficiency, equilibrium strategies and properties.