Engineering
Pulse Compression on IQ: A Smaller FPGA Workload
A short engineering review of decimated-IQ pulse compression for Barker-coded ultrasound, separating filter resources from complete-scanner performance.
By ALEVA | California, USA
Can coded-excitation processing fit a portable ultrasound FPGA without enlarging the device? A useful option is to change where pulse compression runs, rather than simply add compute resources.
The method
In a peer-reviewed January 2026 study, Han, Choi and Yoon perform Barker-code pulse compression after beamformed RF has been demodulated, low-pass filtered and decimated into IQ. The mismatched filter therefore works on a lower-rate representation. Coded excitation and IQ processing themselves are established techniques; the contribution is this implementation and evaluation, not their invention.
The reported prototype uses four-fold decimation and a time-shared filter with 65 coefficients. The authors describe 260 coefficients for the RF-domain alternative. Importantly, only the IQ version was implemented on their FPGA; this is not a measured FPGA-versus-FPGA speed comparison.
The trade-off
The paper reports comparable axial-resolution results under its tested conditions, but also near-field blurring from post-compression errors and sensitivity to frequency-dependent attenuation. Fewer filter coefficients do not establish proportionally faster frames, lower total power or unchanged diagnostic performance.
A proposed OEM acceptance test
Our recommendation is to evaluate the processing boundary, not just the tap count:
Replay identical beamformed input data through the RF reference and IQ candidate; document the demodulator, anti-alias filter and retained bandwidth.
Compare axial profiles and range sidelobes across agreed depths. Include near-field cases rather than selecting only the clearest image.
Record fixed-point settings, FPGA resources and complete processing time, including transfers. Set pass/fail criteria before testing.
Bottom line: decimated-IQ pulse compression is a candidate hardware/software partition for an ultrasound OEM project. Its suitability depends on the signal bandwidth, implementation and acceptance evidence—not a universal speedup claim.
For system context, see ALEVA’s FPGA/GPU architecture guide and transmit/receive boundary guide.
Source
Han M, Choi HJ, Yoon C. A computationally efficient pulse compression method of Barker-coded excitation using a mismatched filter in medical ultrasound imaging. Ultrasonography. January 2026;45(1):30–37. DOI: 10.14366/usg.25135 · Publisher full text.
No figures or clinical images reproduced.