This paper discusses a railway specific cognitive radio that builds upon software defined radio (SDR) platforms to adapt the radio based situational awareness. Cognitive Radio incorporates artificial intelligence based algorithms with reconfigurable software-defined radios that enable automatic adjustments of the radio to improve performance and overcome obstacles the radio may confront in the field (i.e. environmental/man-made interference, occupying the same channel as a user with higher priority, etc.). This paper describes the Railway Cognitive Radio (Rail-CR) architecture and illustrates preliminary results in simulation. The proposed cognitive engine architecture consists of a case-based reasoned (CBR) and a Genetic Algorithm (GA) optimization routine. This paper discusses the overall cognitive architecture, the relationship between the CBR and the GA based on weighted objective functions, and metrics for assessing performance. Methods for case representation, quantifying similarity between cases histories, and techniques for managing case growth rate are presented as well as a proposed test bed SDR platform.