Subject: I need help in my masters thesis -- please read From: venkat <venkathls@xxxxxxxx> Date: Thu, 6 Apr 2006 06:01:43 +0100--0-1372309040-1144299703=:47539 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 8bit dear,, i have some suggestions... With just the Time Delay of arrival of the signal from 2 sensors we just get an arc on which we can localize the sensor under consideration. This is where the need arises for the angle of arrival or direction of arrival(DOA). With this angle we can then use the localization algorithms to determine the position of sensors. Practically the determination of DOA is done by direction of propagation for source emitted signals which is opposite to the direction of arrival. There are a lot of sophisticated methods for estimation of DOA. There are two basic approaches for calculating DOA: Conventional approach [2] and parametric approach [3]. Conventional approach: This approach tries to meet constraints of using minimum number of sources for estimation and also to give an unambiguous estimate without any assumption on the source signal. This method is independent of medium of propagation provided that speed of propagation is constant within sensor array. Parametric approach: In this method conventional beam former and Capon beam former and MUSIC (Multiple Signal Classification) and ESPIRIT (Estimation of signal invariance technique) categorized as subspace-based method are used to estimate DOA. MUSIC can estimate DOA under uncorrelated environment clearly; moreover unitary ESPIRIT can estimate DOA under uncorrelated signal conditions. Unitary ESPIRIT also incorporates forward backward averaging, it conquers problem of coherent signal sources. Conventional approaches for DOA estimation have design simplicity as well as robustness. They also give uncertainty estimate. Parametric methods have high accuracy and efficiency. But the parametric approach becomes unusable under low SNR condition. The inherent inaccuracy is acceptable if we are dealing with broadband signals. But if we are dealing with narrowband signals there has to be a tradeoff between accuracy as well as immunization to noise and signal parameters. Because in narrow band, it is very difficult to determine location as the DOA estimation deteriorates more than exponentially with number of sensors. So it becomes very difficult, rather impossible, to determine the location of the target. We intend to improve the robustness of the parametric approach in low SNR scenarios and design a suitable hardware with suitable interface between the signals and sensors. I feel you use both the models For speech segregation, a recurrent blind separation model (BSS) is tested together with a CASA model, which is based on the localisation cue and the evaluation of the time delay of arrival (TDOA) regards,,, Venkat --------------------------------- Jiyo cricket on Yahoo! India cricket Yahoo! Messenger Mobile Stay in touch with your buddies all the time. --0-1372309040-1144299703=:47539 Content-Type: text/html; charset=iso-8859-1 Content-Transfer-Encoding: 8bit <DIV>dear,,</DIV> <DIV> </DIV> <DIV>i have some suggestions...</DIV> <DIV> </DIV> <DIV class=MsoNormal style="MARGIN: 0in 0in 0pt"><FONT face="Times New Roman" size=3>With just the Time Delay of arrival of the signal from 2 sensors we just get an arc on which we can localize the sensor under consideration. This is where the need arises for the angle of arrival or direction of arrival(DOA). With this angle we can then use the localization algorithms to determine the position of sensors.</FONT></DIV> <DIV class=MsoNormal style="MARGIN: 0in 0in 0pt"><?xml:namespace prefix = o ns = "urn:schemas-microsoft-com:office:office" /><o:p><FONT face="Times New Roman" size=3> </FONT></o:p></DIV> <DIV class=MsoNormal style="MARGIN: 0in 0in 0pt"><FONT size=3><FONT face="Times New Roman">Practically the determination of DOA is done by direction of propagation for source emitted signals which is opposite to the direction of arrival. There are a lot of sophisticated methods for estimation of DOA. There are two basic approaches for calculating DOA: Conventional approach [2] and parametric approach [3].<SPAN style="mso-spacerun: yes"> </SPAN></FONT></FONT></DIV> <DIV class=MsoNormal style="MARGIN: 0in 0in 0pt"><B style="mso-bidi-font-weight: normal"><U><o:p><SPAN style="TEXT-DECORATION: none"><FONT face="Times New Roman" size=3> </FONT></SPAN></o:p></U></B></DIV> <DIV class=MsoNormal style="MARGIN: 0in 0in 0pt"><B style="mso-bidi-font-weight: normal"><FONT face="Times New Roman" size=3>Conventional approach:</FONT></B></DIV> <DIV class=MsoNormal style="MARGIN: 0in 0in 0pt"><FONT face="Times New Roman" size=3>This approach tries to meet constraints of using minimum number of sources for estimation and also to give an unambiguous estimate without any assumption on the source signal. This method is independent of medium of propagation provided that speed of propagation is constant within sensor array. </FONT></DIV> <DIV class=MsoNormal style="MARGIN: 0in 0in 0pt; tab-stops: 48.75pt"><SPAN style="mso-tab-count: 1"><FONT face="Times New Roman" size=3> </FONT></SPAN></DIV> <DIV class=MsoNormal style="MARGIN: 0in 0in 0pt"><B style="mso-bidi-font-weight: normal"><FONT size=3><FONT face="Times New Roman">Parametric approach:<U><SPAN style="mso-spacerun: yes"> </SPAN></U></FONT></FONT></B></DIV> <DIV class=MsoNormal style="MARGIN: 0in 0in 0pt"><FONT size=3><FONT face="Times New Roman"><SPAN style="mso-spacerun: yes"> </SPAN>In this method conventional beam former and Capon beam former and MUSIC (Multiple Signal Classification) and ESPIRIT (Estimation of signal invariance technique) categorized as subspace-based method are used to estimate DOA. MUSIC can estimate DOA under uncorrelated environment clearly; moreover unitary ESPIRIT can estimate DOA under uncorrelated signal conditions. Unitary ESPIRIT also incorporates forward backward averaging, it conquers problem of coherent signal sources. <B style="mso-bidi-font-weight: normal"><o:p></o:p></B></FONT></FONT></DIV> <DIV class=MsoNormal style="MARGIN: 0in 0in 0pt"><B style="mso-bidi-font-weight: normal"><o:p><FONT face="Times New Roman" size=3> </FONT></o:p></B></DIV> <DIV class=MsoNormal style="MARGIN: 0in 0in 0pt"><B style="mso-bidi-font-weight: normal"><o:p><FONT face="Times New Roman" size=3> </FONT></o:p></B></DIV> <DIV class=MsoNormal style="MARGIN: 0in 0in 0pt"><B style="mso-bidi-font-weight: normal"><o:p><FONT face="Times New Roman" size=3> </FONT></o:p></B></DIV> <DIV class=MsoNormal style="MARGIN: 0in 0in 0pt"><B style="mso-bidi-font-weight: normal"><o:p><FONT face="Times New Roman" size=3> </FONT></o:p></B></DIV> <DIV class=MsoNormal style="MARGIN: 0in 0in 0pt"><FONT face="Times New Roman" size=3>Conventional approaches for DOA estimation have design simplicity as well as robustness. They also give uncertainty estimate. Parametric methods have high accuracy and efficiency. But the parametric approach becomes unusable under low SNR condition. The inherent inaccuracy is acceptable if we are dealing with broadband signals. </FONT></DIV> <DIV class=MsoNormal style="MARGIN: 0in 0in 0pt"><FONT face="Times New Roman" size=3>But if we are dealing with narrowband signals there has to be a tradeoff between accuracy as well as immunization to noise and signal parameters. </FONT></DIV> <DIV class=MsoNormal style="MARGIN: 0in 0in 0pt"><FONT face="Times New Roman" size=3>Because in narrow band, it is very difficult to determine location as the DOA estimation deteriorates more than exponentially with number of sensors. So it becomes very difficult, rather impossible, to determine the location of the target.</FONT></DIV> <DIV class=MsoNormal style="MARGIN: 0in 0in 0pt; TEXT-ALIGN: justify; mso-layout-grid-align: none"><FONT face="Times New Roman" size=3>We intend to improve the robustness of the parametric approach in low SNR scenarios and design a suitable hardware with suitable interface between the signals and sensors.</FONT></DIV> <DIV class=MsoNormal style="MARGIN: 0in 0in 0pt; TEXT-ALIGN: justify; mso-layout-grid-align: none"><o:p><FONT face="Times New Roman" size=3> </FONT></o:p></DIV> <DIV class=MsoNormal style="MARGIN: 0in 0in 0pt"><o:p><FONT face="Times New Roman" size=3>I feel you use both the models</FONT></o:p></DIV> <DIV class=MsoNormal style="MARGIN: 0in 0in 0pt"><o:p>For speech segregation, a recurrent blind separation model (BSS) is tested together with a CASA model, which is based on the localisation cue and the evaluation of the time delay of arrival (TDOA)</o:p></DIV> <DIV class=MsoNormal style="MARGIN: 0in 0in 0pt"><o:p></o:p> </DIV> <DIV class=MsoNormal style="MARGIN: 0in 0in 0pt"><o:p>regards,,,</o:p></DIV> <DIV class=MsoNormal style="MARGIN: 0in 0in 0pt"><o:p></o:p> </DIV> <DIV class=MsoNormal style="MARGIN: 0in 0in 0pt"><o:p>Venkat</o:p></DIV><p> <hr size=1> Jiyo cricket on <a href="http://us.rd.yahoo.com/mail/in/mailcricket/*http://in.sports.yahoo.com/cricket/">Yahoo! India cricket</a><br> <a href="http://us.rd.yahoo.com/mail/in/mailmobilemessenger/*http://in.mobile.yahoo.com/new/messenger/">Yahoo! Messenger Mobile</a> Stay in touch with your buddies all the time. --0-1372309040-1144299703=:47539--