Transforming monitoring structures with resilient encoders -- application to repeated games (Q367447)

In this paper the authors propose a monitoring structure of a general repeated game, consisting in (1) \(k\) players who have only partially observed information about the opponents' strategies, and (2) one encoder whose functionality is to send complementary public information to the players. It is assumed that the encoder perfectly observes the players' actions. The paper focuses on two main results: the first one provides the virtually perfect monitoring (VPM) property, which allows the players -- in a best possible way -- to recover the information about the opponents' strategies, and the second result concerns the existence of average cost equilibrium strategies for the players. An application to the repeated prisoner's dilemma illustrates the results.