Ad-hoc networks are independent of any infrastructure. The nodes are autonomous and make their own decisions. They also have limited energy resources. Thus, a node tends to behave selfishly when it is asked to forward the packets of other nodes. Indeed, it would rather choose to reject a forwarding request in order to save its energy. To overcome this problem, the nodes need to be motivated to cooperate. To this end, the authors propose a self-learning repeated game framework to enforce cooperation between the nodes of a network. This framework is inspired by the concept of "The Weakest Link" TV game.