GPU-Accelerated Signal Processing for Passive Bistatic Radar

Round 1

Reviewer 1 Report

This article presented the GPU acceleration method for a passive bistatic radar system. The approach is relatively sound, and the results look promising. The following review points are suggested for further improvement.

1. Please elaborate on how to determine the max range and max Doppler. Can we have both high max range and high max Doppler in this passive radar design through overlapping signal blocks? If the traditional radar range and Doppler dilemma still applies, are there any ways to mitigate the dilemma in the passive radar setup? Please discuss this in the paper.

2. The author chose FM radio waveform. The carrier frequency of FM is low. Therefore, the ambiguous Doppler velocity is high. Also, FM has no cyclic prefix. If a waveform with cyclic prefix is chosen, can the overlapping FFT design still work? Please explain since this paper claims generality of passive waveforms.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

Dear Authors,

The paper explores the realm of passive bistatic radar, a groundbreaking technology that detects targets without actively emitting signals. The focus is on addressing the computational challenges associated with larger data volumes in passive bistatic radar compared to traditional active radiation radar. The research delves into the signal processing flow of passive bistatic radar, proposing a parallel processing scheme under Graphic Processing Unit (GPU) architecture.

The utilization of high-computing-power GPUs and the optimization of algorithms, including Extensive Cancellation Algorithm-Batches (ECA-B), range-Doppler, and constant false alarm detection, demonstrate the paper's commitment to efficient signal processing. The experiments conducted on a passive bistatic radar dataset of natural scenarios showcase the proposed algorithm's ability to achieve a remarkable maximum acceleration ratio of 113.13.

The inclusion of more references could bolster the paper's academic standing. Drawing connections to existing research on passive bistatic radar signal processing, especially those leveraging GPU acceleration, would provide a richer context for the proposed scheme. Additionally, a brief exploration of potential future avenues for research in this domain could enhance the paper's contribution to the field. Here few examples: 1. https://ieeexplore.ieee.org/document/6900723; 2. https://discovery.ucl.ac.uk/id/eprint/10144057/1/IET%20Radar%20Sonar%20%20%20Navi%20-%202022%20-%20Li%20-%20Design%20of%20high%E2%80%90speed%20software%20defined%20radar%20with%20GPU%20accelerator.pdf; 3. https://journals.pan.pl/Content/87923/PDF/47.pdf; 4. https://www.researchgate.net/publication/238524207_Combined_use_of_graphics_processing_unit_GPU_and_Central_Processing_Unit_CPU_for_passive_radar_signal_data_elaboration; 5. https://us.artechhouse.com/Signal-Processing-for-Passive-Bistatic-Radar-P2044.aspx

The conclusions, while promising, could benefit from a more nuanced discussion of the limitations of the proposed scheme. Addressing potential challenges or scenarios where the acceleration ratio might vary could provide a more comprehensive understanding for readers and researchers interested in implementing similar techniques.

In summary, the paper makes a valuable contribution to the field of passive bistatic radar signal processing, leveraging GPU acceleration for enhanced efficiency. Minor revisions, including the addition of relevant references and a more nuanced discussion of conclusions, would further strengthen the paper's impact and appeal to a broader audience of researchers in the radar technology domain.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 3 Report

Regarding section 4.2:
- Did you use data containing a real target or did you add a simulated target to the data?
- If these are real targets, can you indicate what kind of target they are?
- Figure 7 a) and b) seem to show an aerial target with a credible Doppler corresponding to a target moving away from the radar. On the other hand, c) and d) shows a zero Doppler target, which is a rather singular case for an aerial target. Can you comment on this point?
- It would be useful to add a Z scale so that we have a precise view of the levels respect to the background noise, which would be useful for making a precise judgment on what you are proposing.
- It would also be useful to present the intermediate results before and after the removal of the direct signal (eq. 1 and 2)

Author Response

Please see the attachment

Author Response File: Author Response.pdf