Skip to main content
Log in

Exponential-Reproducing-Kernel-Based Sparse Sampling Method for Finite Rate of Innovation Signal with Arbitrary Pulse Echo Position

  • Published:
Circuits, Systems, and Signal Processing Aims and scope Submit manuscript

Abstract

An exponential reproducing kernel (ERK) has been applied to data sparse sampling for finite rate of innovation (FRI) signals with the characteristic of flexible parameter setting in contrast to other sampling kernels. However, the signal reconstruction process may fail if a pulse echo is present in certain positions. To solve this problem, a novel ERK sparse sampling method was developed for arbitrary pulse echo positions. A constraint relationship between the pulse echo position and reproduced-exponent area was deduced, revealing the cause of invalid signal reconstruction from sparse sampling data. A new sampling time interval calculation algorithm is presented in this paper. Through the proposed method, FRI signals can be reconstructed accurately with a pulse echo present at arbitrary positions without increasing the quantity of sparse sampling data. The signal reconstruction effectiveness and accuracy were verified through simulation experiments. The proposed technique can be used to improve the flexibility of the ERK sparse sampling method in actual applications.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6

Similar content being viewed by others

References

  1. J. Chen, Y.T. Gao, C.H. Ma, Y.H. Kuo, Compressive sensing image reconstruction based on multiple regulation constraints. Circuits Syst. Signal Process. 36(4), 1621–1638 (2017)

    Article  Google Scholar 

  2. M. Chen, F. Ding, L. Xu, T. Hayat, A. Alsaedi, Iterative identification algorithms for bilinear-in-parameter systems with autoregressive moving average noise. J. Frankl. Inst. 354(17), 7885–7898 (2017)

    Article  MathSciNet  MATH  Google Scholar 

  3. P.L. Dragotti, M. Vetterli, T. Blu, Sampling moments and reconstructing signals of finite rate of innovation: Shannon meets strange-fix. IEEE Trans. Signal Process. 55(5), 1741–1757 (2007)

    Article  MathSciNet  MATH  Google Scholar 

  4. F. Ding, X. Wang, Q. Chen, Y. Xiao, Recursive least squares parameter estimation for a class of output nonlinear systems based on the model decomposition. Circuits Syst. Signal Process. 35(9), 3323–3338 (2016)

    Article  MathSciNet  MATH  Google Scholar 

  5. F. Ding, Y. Wang, J. Dai, Q. Li, Q. Chen, A recursive least squares parameter estimation algorithm for output nonlinear autoregressive systems using the input–output data filtering. J. Frankl. Inst. 354(15), 6938–6955 (2017)

    Article  MathSciNet  MATH  Google Scholar 

  6. K. Gedalyahu, R. Tur, Y.C. Eldar, Multichannel sampling of pulse streams at the rate of innovation. IEEE Trans. Signal Process. 59(4), 1491–1504 (2010)

    Article  MathSciNet  MATH  Google Scholar 

  7. G.H. Golub, P. Milanfar, J. Varah, A stable numerical method for inverting shape from moments. Soc. Ind. Appl. Math. 21(4), 1222–1243 (1999)

    MathSciNet  MATH  Google Scholar 

  8. G. Itzhak, E. Baransky, N. Wagner: A hardware prototype for sub-Nyquist radar sensing, in International Itg Conference on Systems (2013), pp. 1–6

  9. S.P. Song, Y.H. Shao, Y. Du, Survey of sampling methods. J. Data Acquis. Process. 31(3), 452–463 (2016)

    Google Scholar 

  10. E.V. Strelkova, V.T. Shevaldin, Local exponential splines with arbitrary knots. Proc. Steklov Inst. Math. 288(1), 189–194 (2015)

    Article  MathSciNet  MATH  Google Scholar 

  11. R. Tur, Y.C. Eldar, Z. Friedman, Innovation rate sampling of pulse streams with application to ultrasound imaging. IEEE Trans. Signal Process. 59(4), 1827–1842 (2011)

    Article  MathSciNet  MATH  Google Scholar 

  12. J.A. Urigüen, T. Blu, P.L. Dragotti, FRI sampling with arbitrary kernels. IEEE Trans. Signal Process. 61(21), 5310–5323 (2013)

    Article  Google Scholar 

  13. M. Unser, T. Blu, Cardinal exponential splines: part I—theory and filtering algorithms. IEEE Trans. Signal Process. 53(4), 1425–1438 (2005)

    Article  MathSciNet  MATH  Google Scholar 

  14. M. Vetterli, P. Marziliano, T. Blu, Sampling signals with finite rate of innovation. IEEE Trans. Signal Process. 50(6), 1417–1428 (2002)

    Article  MathSciNet  MATH  Google Scholar 

  15. Y.J. Wang, Sampling complex pulse streams with finite rate of innovation methods. J. Electron. Inf. Technol. 35(7), 1606–1611 (2013)

    Article  Google Scholar 

  16. Y.J. Wang, M. Li, G.F. Liu, Sampling complex pulse streams with finite rate of innovation methods. J. Electron. Inf. Technol. 35(7), 1606–1611 (2013)

    Article  Google Scholar 

  17. L. Xu, F. Ding, Recursive least squares and multi-innovation stochastic gradient parameter estimation methods for signal modeling. Circuits Syst. Signal Process. 36(4), 1735–1753 (2017)

    Article  MATH  Google Scholar 

  18. L. Xu, F. Ding, Iterative parameter estimation for signal models based on measured data. Circits Syst. Signal Process. 37(7), 3046–3069 (2018)

    Article  MathSciNet  Google Scholar 

  19. G. Yang, The property of a special type of exponential spline function. Adv. Pure Math. 5(13), 804–807 (2015)

    Article  Google Scholar 

Download references

Acknowledgements

This project is supported by the National Natural Science Foundation of China (Grant No. 51375217). This manuscript has been edited by Mark Darvill at Wallace Academic Editing and is considered to be improved in grammar and native English usage.

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Shoupeng Song.

Appendix: Concept of RE-area

Appendix: Concept of RE-area

The exponential reproducing formula of the \( M \)th-order E-spline function \( \beta_{{\vec{\alpha }}} \left( t \right) \) with \( \beta_{{\vec{\alpha }}} \left( 0 \right) = 0 \), \( \vec{\alpha } = \alpha_{1} ,\alpha_{2} , \ldots ,\alpha_{M} \), can be expressed as [10, 13, 19]

$$ \mathop \sum \limits_{{k \in {\mathbf{Z}}}} c_{m,k} \beta_{{\vec{\alpha }}} \left( {t - k} \right) = e^{{\alpha_{m} t}} , \quad m = 1,2, \ldots ,M $$
(14)

where \( c_{m,k} \) denotes the corresponding exponential reproducing weight coefficient.

Generally, the exponent \( e^{{\alpha_{m} t}} \) can be produced by summing the weighted terms of \( \beta_{{\vec{\alpha }}} \left( {t - k} \right),k \in {\mathbf{Z}} \) in an infinite time shift set. As shown in Fig. 7, in a finite time shift set \( \varvec{K}_{f} = \left\{ {k |k \in \left[ {K_{1} ,K_{2} } \right], k \in {\mathbf{Z}}} \right\} \), the exponential reproducing formula is expressed as

$$ r\left( t \right) = \mathop \sum \limits_{{k \in \varvec{K}_{f} }} c_{m,k} \beta_{{\vec{\alpha }}} \left( {t - k} \right) $$
(15)

where the time support of the reproduced function shifts to \( \left[ {K_{1} ,K_{2} + M} \right). \)

Fig. 7
figure 7

Sketch map of reproducing exponent

Clearly, the exponent \( e^{{\alpha_{m} t}} \) cannot be generated in the whole time support of \( r\left( t \right). \) The time support area of the weighted terms \( c_{{m,K_{1} - 1}} \beta_{{\vec{\alpha }}} \left( {t - K_{1} + 1} \right) \) and \( c_{{m,K_{2} + 1}} \beta_{{\vec{\alpha }}} \left( {t - K_{2} - 1} \right) \) must be deleted to reproduce the exponent, or in other words, the areas \( \left( {K_{1} - 1,K_{1} + M - 1} \right) \) and \( \left( {K_{2} + 1,K_{2} + M + 1} \right) \), respectively. Therefore, when \( K_{2} - K_{1} \ge M - 2 \) and \( M \ge 2 \) are both satisfied,

$$ r\left( t \right)\left\{ {\begin{array}{*{20}c} { = e^{{\alpha_{m} t}} ,} & { t \in \varvec{E}} \\ { \ne e^{{\alpha_{m} t}} ,} & {t \in {\complement }_{{\left[ {K_{1} ,K_{2} + M} \right)}} \varvec{E}} \\ \end{array} } \right. $$
(16)

where \( {\complement }_{\text{U}} {\text{A}} \) denotes the supplementary set of set \( {\text{A}} \) in set \( {\text{U}} \) and the time interval \( \varvec{E} \) is defined as the RE-area of the \( M \) th-order E-spline function \( \beta_{{\vec{\alpha }}} \left( t \right) \) under the time shift set \( \varvec{K}_{f} \) and is expressed as

$$ \varvec{E} = \left[ {K_{1} + M - 1,K_{2} + 1} \right] $$
(17)

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Song, S., Shen, J. Exponential-Reproducing-Kernel-Based Sparse Sampling Method for Finite Rate of Innovation Signal with Arbitrary Pulse Echo Position. Circuits Syst Signal Process 38, 1179–1193 (2019). https://doi.org/10.1007/s00034-018-0903-8

Download citation

  • Received:

  • Revised:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s00034-018-0903-8

Keywords

Navigation