BEIHANG UNIVERSITY 201 LAB

1. IGARSS 2011 PARALLEL FREQUENCY RADAR VIA COMPRESSIVE SENSING • You Yanan, Li Chunsheng, Yu Ze BEIHANG UNIVERSITY 201 LAB

2. OUTLINE 1. INTRODUCTION 2. background of compressive sensing 3. PARALLEL FREQUENCY RADAR BASED ON COMPRESSIVE SENSING 4. simulation results 5. CONCLUSION

3. 1. INTRODUCTION • Traditional radar utilizes Shannon-Nyquist theorem for high bandwidth signal sampling, which induces the complicated system. • Compressive sensing (CS) indicates that the compressible signal using a few measurements can be reconstructed by solving a convex optimization problem.

4. 1. INTRODUCTION • The structure of compressive sensing radar with frequencies transmitted in parallel is firstly presented in On compressive sensing applied to radar • Parallel frequency radar theoretically cannot degrade the resolution compared with a traditional radar system and effectively reduces the sampling rate.

5. 2. BACKGROUND OF COMPRESSIVE SENSING sparse scene S convex problem • Basis Pursuit (BP) , Matching Pursuit (MP) and Orthogonal Matching • Pursuit (OMP) can solve this convex problem • Sparse scene S can be reconstructed by the above algorithms.

6. 3. PARALLEL FREQUENCY RADAR BASED ON • COMPRESSIVE SENSING 3.1 Observation mode • The radar platform moves along track (azimuth) direction with constant velocity. • two-dimension observation scene is limited with the along track (azimuth) direction and the cross track (range) direction. • Parallel frequency radar synchronously transmits M single frequency pulses in each azimuth and then received the phase shift echoes.

7. 3. PARALLEL FREQUENCY RADAR BASED ON • COMPRESSIVE SENSING 3.2 Resolutionand sampling rates Ground observation grid Azimuth Range • Range resolution of traditional pulse compressive radar • Parallel frequency radar synchronously transmits M single frequency pulses in each azimuth with the constant . • Range resolution of parallel frequency radar via compressive sensing Conclusion ：If we set the appropriate M and ≈

8. 3. PARALLEL FREQUENCY RADAR BASED ON • COMPRESSIVE SENSING 3.2 Resolutionand sampling rates Conclusion：If we set the appropriate M and D The sampling rate of parellel frequency radar is M/T complex samples per second. Define the time-bandwidth product to be D=BT, B is the total bandwidth of M single frequency pulses. Then, the sampling rate is shown as BM/D. ＞ BM/D B Traditional radar system Parallel frequency radar system

9. 3. PARALLEL FREQUENCY RADAR BASED ON • COMPRESSIVE SENSING 3.3 Imaging based on compressive sensing • The phases of echo signal contain the information of the targets. The kth echo in a certain azimuth is The phase information of the target is The discrete representation is Yis identity matrix

10. 3. PARALLEL FREQUENCY RADAR BASED ON • COMPRESSIVE SENSING Ka estimation • The targets will be focused on the respective zero Doppler by Ka. • Ka is calculated precisely before compressing the azimuth phases. • Ka varies with the range cells and the signal frequencies.

11. 3. PARALLEL FREQUENCY RADAR BASED ON • COMPRESSIVE SENSING F construction 1D observation grid Mechoes phase detector Msingle frequency pulses M frequencies 3D prior F PriorF Measurement matrix is constructed by prior phase information using discrete grids.

12. 3. PARALLEL FREQUENCY RADAR BASED ON • COMPRESSIVE SENSING Observation scene reconstruction Observation scene Observation grids Mechoes phase detector M frequencies Msingle frequency pulses Na×Nr PriorF recover 1×Nr OMP M×Nr M frequencies

13. 3. PARALLEL FREQUENCY RADAR BASED ON • COMPRESSIVE SENSING Illation Echo equation 1 Instantaneous range Approximate representation Echo equation 2 Demodulating and detecting the phase shift

14. 4. simulation results The parameters in simulations: M = 30, △f= 5MHz, RMAX = 20km andN = 512. Azimuth cross section(amplitude) Imaging result of point target of the parellel frequency radar via compressive sensing Range cross section(amplitude)

15. 4. simulation results Imaging results of pulse compression radar and parallel frequency radar via compressive sensing with single or several point targets Range Range Compressive sensing algorithm of parallel frequency radar RD algorithm of pulse compression radar Parallel frequency radar via compressive sensing Pulse compression radar

16. 4. simulation results Imaging results of traditional SAR and parallel frequency radar via compressive sensing with the same data amount Parallel frequency radar via compressive sensing Traditional SAR

17. Amplitude Amplitude Range Azimuth Azimuth Range Azimuth Range 4. simulation results Parallel frequency radar via compressive sensing Pulse compression radar Azimuth Range Conclusion: Sampling rate is reduced by narrow-bandwidth of single frequency pulse. The sparse scene is reconstructed using 10% original data. Side lobes of the range disappear via compressive sensing.

18. 4. simulation results The ability of distinguishing the adjacent targets in the range direction Pulse compression radar Parallel frequency radar via compressive sensing

19. 5. CONCLUSION • This presentation introduces parallel frequency radar and imaging approach based on compressive sensing. • The novel radar can reduce data rate and maintain range resolution with the premise of the appropriate parameters. • The measurement matrix F is composed of the priori phases. The observation scene is reconstructed by OMP.

20. 5. CONCLUSION • The imaging approach based on compressive sensing avoids side lobes of the range. • The noise in received signal is not taken into account in the simulations, it is not ignored in practice. • More researches are required on the capability resisting the noise interference of the signal processing based on compressive sensing in the future.

21. IGARSS 2011 Thank you for your attention! BEIHANG UNIVERSITY 201 LAB