A controller in adaptive control theory is a critical part in mission critical applications in military and computer-controlled systems. An ability to identify and follow the binary instruction execution in the controller part enables fault identification and malware detection in safety critical applications. Electromagnetic field emission based identification of controllers execution state from distance will help ascertain security vulnerabilities early on. machine learning models for instruction identification, Principal Component Analysis (PCA), Adaptive Boosting (AB) and Naïve Bayes (NB) were developed to meet this goal. Our preliminary results of implementation on a 2-stage pipelined controller processor architecture demonstrate that the EM side-channel classification approach identifies a controller execution state in Adaptive control with 93% success rate.