Frequency-Selective Robust Detection and Estimation of Multiple-Polymorph QR Signals

Research output: Contribution to journalArticle


Nuclear quadrupole resonance (NQR) is a non-invasive, solid state, radio frequency (RF) technique, able to distinguish between polymorphic forms of certain compounds. Exploiting the signals from multiple polymorphs is important in explosives detection, whilst quantifying these polymorphs is important in pharmaceutical applications. Recently proposed hybrid algorithms, able to process the signals from multiple polymorphs, assume that the amplitudes associated with each polymorph are known to be within a scaling. Any error in this a priori information will lead to performance degradation in these algorithms. In this paper, we develop a robust hybrid algorithm allowing for uncertainties in the assumed amplitudes, extending a recently proposed robust algorithm, formulated for single polymorphs, to process signals from multiple polymorphs. In the proposed robust algorithm, the amplitudes are allowed to vary within an uncertainty hyper-sphere whose radius is evaluated using analytical expressions derived herein. Extensive numerical investigations indicate that the proposed algorithm provides significant performance gains as compared to both the existing hybrid algorithms, when uncertainties in the amplitudes exist, and the existing robust algorithm, when there are multiple polymorphs. Finally, the Cramér–Rao lower bound is derived for the uncertain data case as a reference for the quantification problem.


External organisations
  • External Organization - Unknown
Research areas and keywords

Subject classification (UKÄ) – MANDATORY

  • Probability Theory and Statistics


  • Nuclear quadrupole resonance, Constrained least squares, Explosives detection
Original languageEnglish
Pages (from-to)834-843
JournalSignal Processing
Issue number4
Publication statusPublished - 2008
Publication categoryResearch
Externally publishedYes