In RF demodulation pipelines, system failure is often characterised by the emergence of
ghost modes and a collapse of true detections. These “failure rims” occupy narrow regions
of a high-dimensional parameter space spanned by signal-to-noise ratio (SNR), frequency offset ∆f, amplitude modulation depth (AM) and the Q factor of the analog front end. Mapping such rims with exhaustive sweeps is prohibitively expensive. This paper demonstrates
how uncertainty-seeking acquisition coupled with surrogate models can be used to chart failure
boundaries efficiently. A synthetic benchmark illustrates failure rims in three parameter planes:
SNR × ∆f, SNR × AM, and ∆f × Q. Contour plots reveal how ghost hit rates rise and true
hit rates collapse along distinct ridges, providing actionable insights for tuning RF pipelines.