For FMCW fast chirping signal, the maximum unambiguous velocity is Vmax = wavelength/4Tc, with Tc is the chirp duration. To get high unambiguous velocity, chirp duration becomes too small to be practical in real system. Several possible solutions to extend maximum unambiguous velocity are:
1) Velocity disambiguation by varying inter-chirp time
One of the techniques discussed in literature to resolve ambiguities is based on varying the pulse repetitions frequency and using the resulting movement of the detected (‘aliased’) bins to infer the true velocity. Then use Chinese-Remainder-Theorem (CRT, en.wikipedia.org/.../Chinese_remainder_theorem) theory to solve the ambiguity. Please refer to the following two papers for details.
Reference:
“Ambiguity elimination in HF FMCW radar systems”, M.Musa and SSalous, 2000
“Radar target detection and Doppler ambiguity resolution”, Matthias Kronauge, etc, 2010
User can also change inter-chirp starting frequency or other chirp parameters to achieve maximum velocity extension
2) Velocity disambiguation based on observation of successive range estimates
Range measurements for an object that have been obtained across successive consecutive frames can be used to disambiguate velocity. Such measurements are free from velocity aliasing, but with much coarser accuracy. However ‘coarse’ velocity estimates obtained from range measurements can be combined with the more accurate (but possibly aliased) velocity estimates obtained from the Doppler-FFT to get the final accurate velocity measurement.
In general the procedures are:
– Associate points across consecutive frames
– Estimate the velocity based on range change and frame time. R1 and R2 are the range of the point across two frames. Tc is the frame duration. The velocity can be estimated as with coarse accuracy V_coarse = (R1-R2)/Tc
– Estimate ambiguity integer n from the coarse velocity estimation, n_est = round(V_coarse – V_fine/2Vmax)where V_fine is fine velocity estimation from Doppler shift within a frame.
– The final true velocity is V_true = V_fine + 2 x n_est x Vmax
To achieve accurate estimation of n_est, associating multiple frames (more than 2) is necessary, which leads to extra latency caused by multiple frame association.
To achieve accurate estimation of , associating multiple frames
