In 2014, a measurement campaign was made at CEA Cadarache using different versions of a high sensibility miniature fission chamber prototype with two fissile layers. The current project studies the possibilities of discrimination between signals coming from a fission in the fissile layer covering the anode or in the fissile layer covering the cathode. A specific fission chamber simulation tool, named Trace, has been developed. Simulation results show the feasibility of discrimination based on pulse shape. A machine learning classification al- gorithm is applied to the available measurement data. Misleading discrimination parameters are identified and the limits of the possible use of the current data for discrimination pur- poses are highlighted. A promising criterium based on the pulse convexity is presented, and recommendations for the development of a future optimized detector are given.