Frequency domain sequence design algorithm minimising weighted integrated sidelobe level

Biskin O. T., AKAY O.

IET RADAR SONAR AND NAVIGATION, vol.13, no.3, pp.464-472, 2019 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 13 Issue: 3
  • Publication Date: 2019
  • Doi Number: 10.1049/iet-rsn.2018.5321
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus
  • Page Numbers: pp.464-472
  • Keywords: frequency-domain analysis, sequences, correlation methods, minimisation, iterative methods, computational complexity, low-autocorrelation sidelobes, predetermined autocorrelation sidelobes, majorisation minimisation method, frequency domain representation, WISL metric, iterative algorithm, cyclic algorithms, time domain MM-based counterparts, frequency domain sequence design algorithm, closed-form solution, acceleration scheme, WAVE-FORM SYNTHESIS, BINARY SEQUENCES, OPTIMIZATION METHODS, AUTOCORRELATION, CODES, SEARCH
  • Dokuz Eylül University Affiliated: Yes


Unimodular sequences with low-autocorrelation sidelobes are employed in various applications of radar and communication systems. Suppressing some predetermined autocorrelation sidelobes of a designed sequence can be accomplished by performing minimisation of the metric of weighted integrated sidelobe level (WISL). In this study, the authors propose a new algorithm to design unimodular sequences utilising the majorisation minimisation (MM) method for directly minimising WISL in the frequency domain. The proposed algorithm allows the design of long sequences in a computationally efficient manner. As the first step of the algorithm, a function majorising the frequency domain representation of the WISL metric is introduced. Then, a closed-form solution for the minimisation of WISL is derived followed by its realisation as an iterative algorithm. They also provide an acceleration scheme to allow efficient implementation of the designed algorithm. Numerical experiments show that the proposed algorithm not only outperforms existing cyclic algorithms with respect to computation time and suppression of desired autocorrelation sidelobes but also converges in less number of iterations than its time domain MM-based counterparts.