- Nowadays, wireless channel emulators are designed for channel models that require the evaluation of logarithmic, trigonometric, exponential, and other transcendental functions. These channel emulators are used for testing wireless communication standards associated to Weibull, Suzuki, Nakagami, Rayleigh and Gaussian distributions. The piecewise polynomial approximation (PPA) technique allows the evaluation of functions with an accuracy level that depends on the used segmentation method; being the uniform and non-uniform by the power of two, the most commonly employed. However, these segmentation techniques lack the required flexibility to be useful for any function since the signal to quantization noise ratio (SQNR) is highly dependent on the function at hand. A new function segmentation methodology is presented based on the first and second derivative. Simulation results show significant SQNR advantages when the proposed method is compared with state-of-art segmentation techniques. © 2016 IEEE.