A Method to Derive Local Interaction Strategies for Improving Cooperation in Self-Organizing Systems

To achieve a preferred global behavior of self-organizing systems, suitable local interaction strategies have to be found. In general, this is a non-trivial task. In this paper, a general method is proposed that allows to systematically derive local interaction strategies by specifying the preferred global behavior. In addition, the resulting strategies can be evaluated using Markovian analysis. Then, by applying the proposed method exemplarily to the iterated prisoner’s dilemma, we are able to systematically generate a cooperation-fostering strategy which can be shown to behave similar to the “tit for tat with forgiveness” strategy that, under certain circumstances, outperforms the well-known “tit for tat” strategy used, for instance, in BitTorrent peer-to-peer file-sharing networks.