A distributed M2M local cloud platform which consists of distributed M2M gateways, needs to be equipped with an Intrusion Detection System (IDS) to monitor its resources against security attacks, especially from the insider, e.g. another gateway within the local cloud. In this paper, the interaction between rational attacker and defender in the context of an M2M local cloud platform as a multi-stage Bayesian game is studied. In this game formulation, a defender is able to update its belief upon the maliciousness of the attacker. The feasible Nash equilibrium of the game is reviewed and an analytical framework for the rational attacker and defender is provided for a given set of resources with different security values under some constraints on the attack and monitor resources. In the numerical analysis, it can be shown that by having multiple resources to be attacked and/or monitored simultaneously provides a kind of diversity which helps to improve the belief update of the defender.