The MRF framework has been recently investigated to estimate T1, T2 and apparent diffusion coefficient (ADC) from a single scan with b-tensor encoding schemes. However, the current human-designed experiment protocol for multi-dimensional MRF is still subject to limited measurement accuracy. Here we propose to adapt the Cramer-Rao Bounds to optimize multi-dimensional MRF scan for simultaneous quantification of relaxation and diffusion. The optimization framework explores all possible combinations of MRF sequence parameters, including flip angles, TRs, b-values, and preparation modules to seek the optimal tissue parameter encoding scheme. The optimized sequence simultaneously improves the measurement precision of T1, T2 and ADC.