This paper analyses the literature to identify ideas which may be applied to decentralized real-time railway traffic management. This system represents a new way for dealing with railway traffic perturbations in absence of a central decision maker. Specifically, we are interested in identifying techniques that may constitute suitable automatic mechanisms for the emergence of an effective system behaviour. In this literature review, we discuss the possibility of exploiting the existing research works on other transport modes. The analysis of these works makes it clear that real-time railway traffic management is very peculiar. Hence, we consider different approaches: hierarchical self organization, task allocation, reinforcement learning, consensus, auction and coopetition techniques. Some promising possibilities emerge, which we analyse proposing ideas for modelling decentralized real-time railway traffic management.