Recently, I study the MUSIC doa in the circular harmonic domain, but I found the estimation results always have a big bias in some specific angles(angles near to 0 or 180 degree), as showing in the Figureenter image description here, I'm so confused because the estimation deviation in other angles is small, can someone told me the possible reason?

  • $\begingroup$ Do you find these results consistent for all data sets? $\endgroup$ Jun 22, 2020 at 3:20
  • $\begingroup$ Yes, I set different T60/SNR(This figure‘s setting isT60=200ms, SNR=15dB), and the tendency is similar(the bias of estimation of the sources near to 0 or 180 degrees is large).Ps: the size of circles define the RMSE of estimation(which is amplified by a constant factor to show clearly) $\endgroup$
    – Jeff Wang
    Jun 23, 2020 at 4:11
  • $\begingroup$ What estimation algo have you deployed? That could let us know further on why we see this behavior $\endgroup$ Jun 23, 2020 at 7:07
  • $\begingroup$ I used Circular Harmonic domain MUCIC(Transfer element domain data into circular harmonic domain and then applying broadband CSS MUSIC) $\endgroup$
    – Jeff Wang
    Jun 24, 2020 at 8:06


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.