/FFT_Error_Detection_Energy_Saving

Algorithmic error detection within Fourier context to enable low power-energy solutions based on undervolting

Primary LanguageMATLAB

Fault injection can not be implemented simply by adding noise. The MAC units within FFT produce error depending on the input data. A data that practices the longest delay paths is more likely to generate an error. Take this as an example, adding 0000 and 0000 has the lowest error potential and adding 1111 and 1111 has the highes.

alt text