It looks like the first surface of your lens refracts rays at a sufficiently high-NA that the Gaussian waist of the pilot beam becomes very small, less than the wavelength, hence the warning you are seeing.  The small aspherical structure on the second lens surface, which acts to reduce the focused beam NA, is definitely unique.  I think a more common approach would be to reduce the NA of the first surface, keep the second surface planar, and place the fiber farther away.  In any event, I assume you have a specific reason for keeping the fiber close to the lens output surface. 

Instead of using POP, you might want to try the built-in single-mode fiber coupling tool.  Here’s a version that mimics your POP setup:

In POP, you are using wavelength 2 (656 nm) and the fiber beam waist is 2.22 um, which corresponds to an NA = 0.94 -- so I’ve used those values in the fiber coupling analysis.  The “Huygens Integral” option allows for the use of higher-NA beams compared to the FFT PSF or POP technique that may begin breaking down.  You can use the FICL merit function operand to report the coupling value and/or to help with optimization.