If you have access to VerilogA I believe you can convolve two time functions using a pwl source and a z-domain filter. The pwl source pulls in one time function and drives the z-domain filter. The z-domain filter is an FIR filter with the taps equal to the time-domain samples of the second time function. The only thing not ideal about this approach is the first order hold introduced by the pwl source but if the sampling rate is high enough that should be irrelevant.

If you do not have access to z-domain block, I guess you could fit a filter to the Fourier transform of second time function and replace the z-domain block with that filter.