Zero Padding

Zero padding improves range accuracy by increasing the FFT input length, which provides finer frequency (range) bin spacing.
It does not add new information, but enables interpolation of the FFT spectrum for more precise peak location estimation.

 

 

 

 

 

1. Signal Flow and Processing

For sawtooth chirps, only the up-chirp in Zone2 is valid, while Zone1, Zone3 and Zone4 regions are zero-padded;

for triangular chirps, both Zone2 and Zone3 are valid and Zone1 and Zone4 are zero-padded.

 

 

 

 

 

 

2. Example Cases

Increasing zero padding is effective up to about 2×N–4×N,  beyond which additional padding provides negligible improvement;

here, N is the number of valid data samples used for the FFT.

 

Zero Padding Factor	Effect
1×N	No zero padding, strong FFT bin quantization
2×N	Practical benefit begins, most bin quantization reduced
4×N	Largely sufficient, stable peak location estimation
8×N or more	Only visual smoothness improves, no accuracy gain

 

 

This is an example comparing the Range FFT spectrum of a 24 GHz FMCW radar for a virtual target at 5 m, with and without zero padding,

following the concept that Zone2 contains the valid data while Zone1, 3 and Zone3 are zero-padded before the FFT.

 

 

 

Condition of an example graph

Item	Value
Bandwidth (B)	200 MHz
Theoretical range resolution	ΔR = c / (2B) ≈ 0.75 m
Chirp time (Tc)	20 ms
Chirp slope (S)	B / Tc
ADC sampling rate (Fs)	20 kHz
Valid samples (Zone2, N)	256
Target range (R)	5 m
Beat frequency (fb)	fb = 2SR / c

 

Processing and Effects

Item	No Zero Padding	With Zero Padding
FFT size (NFFT)	256	2048
Zero padding factor	1×	8×
Peak range location	~4.68 m	~4.98 m
Range bin spacing	~1.17 m/bin	~0.146 m/bin
Practical effect	Strong bin quantization	Reduced quantization, smoother spectrum