What is the best way to distinguish the phase modulated (say pi/4 dqpsk) signal from noise in baseband? I need a function in radio to scan through different carrier frequencies and reject the frequencies which does not have a modulated signal. Any suggestion for an efficient algorithm?
Method to distinguish a phase modulated signal from noise signal
Started by ●December 16, 2009
Reply by ●December 16, 20092009-12-16
rseb wrote:> What is the best way to distinguish the phase modulated (say pi/4 dqpsk) > signal from noise in baseband? I need a function in radio to scan through > different carrier frequencies and reject the frequencies which does not > have a modulated signal. Any suggestion for an efficient algorithm?Demodulate the signal incoherently, compute RMS of error vector, compare with a threshold. You might want separate thresholds for phase and amplitude of error. Vladimir Vassilevsky DSP and Mixed Signal Design Consultant http://www.abvolt.com
Reply by ●December 16, 20092009-12-16
On Dec 16, 7:57�am, "rseb" <reny.sebast...@gmail.com> wrote:> What is the best way to distinguish the phase modulated (say pi/4 dqpsk) > signal from noise in baseband? I need a function in radio to scan through > different carrier frequencies and reject the frequencies which does not > have a modulated signal. Any suggestion for an efficient algorithm?Raise the baseband signal to an integer power to collapse the modulation into CW. Then use an FFT or other technique to identify the CW stick. It won't be there if the input is noise. John
Reply by ●December 16, 20092009-12-16
John wrote:> On Dec 16, 7:57 am, "rseb" <reny.sebast...@gmail.com> wrote: > >>What is the best way to distinguish the phase modulated (say pi/4 dqpsk) >>signal from noise in baseband? I need a function in radio to scan through >>different carrier frequencies and reject the frequencies which does not >>have a modulated signal. Any suggestion for an efficient algorithm? > > > Raise the baseband signal to an integer power to collapse the > modulation into CW. Then use an FFT or other technique to identify the > CW stick. It won't be there if the input is noise.Problem with this it is not very cheap to distinguish CW sticking out of the background. Vladimir Vassilevsky DSP and Mixed Signal Design Consultant http://www.abvolt.com
Reply by ●December 16, 20092009-12-16
On Wed, 16 Dec 2009 06:57:16 -0600, rseb wrote:> What is the best way to distinguish the phase modulated (say pi/4 dqpsk) > signal from noise in baseband? I need a function in radio to scan > through different carrier frequencies and reject the frequencies which > does not have a modulated signal. Any suggestion for an efficient > algorithm?Look for the carrier? -- www.wescottdesign.com
Reply by ●December 16, 20092009-12-16
On 12/16/2009 8:46 AM, John wrote:> On Dec 16, 7:57 am, "rseb"<reny.sebast...@gmail.com> wrote: >> What is the best way to distinguish the phase modulated (say pi/4 dqpsk) >> signal from noise in baseband? I need a function in radio to scan through >> different carrier frequencies and reject the frequencies which does not >> have a modulated signal. Any suggestion for an efficient algorithm?How fast does it have to work? If the demod has reasonably fast acquisition and you can take some time scanning, then it may be okay to just step through channels and see whether you get lock or not. Sometimes just detecting symbol clock lock is good enough, and that happens fairly quickly.> Raise the baseband signal to an integer power to collapse the > modulation into CW. Then use an FFT or other technique to identify the > CW stick. It won't be there if the input is noise. > > JohnThis tends to not work well at low SNR, since the integer exponent amplifies the noise quite a bit. If the operating range is only ever higher SNR, it works well enough, but in those conditions other, simpler, techniques work well, too. -- Eric Jacobsen Minister of Algorithms Abineau Communications http://www.abineau.com
Reply by ●December 16, 20092009-12-16
Tim Wescott wrote:> On Wed, 16 Dec 2009 06:57:16 -0600, rseb wrote: > >> What is the best way to distinguish the phase modulated (say pi/4 dqpsk) >> signal from noise in baseband? I need a function in radio to scan >> through different carrier frequencies and reject the frequencies which >> does not have a modulated signal. Any suggestion for an efficient >> algorithm? > > Look for the carrier?Does anybody remember the original Panadapter of late '40s vintage? Jerry -- Engineering is the art of making what you want from things you can get. ¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
Reply by ●December 16, 20092009-12-16
>Raise the baseband signal to an integer power to collapse the >modulation into CW. Then use an FFT or other technique to identify the >CW stick. It won't be there if the input is noise. > >John >I never heard the term "CW Stick". Cant get any reference online too. It will be helpful if you could explain a bit more on this.
Reply by ●December 16, 20092009-12-16
On Dec 16, 2:48�pm, "rseb" <reny.sebast...@gmail.com> wrote:> >Raise the baseband signal to an integer power to collapse the > >modulation into CW. Then use an FFT or other technique to identify the > >CW stick. It won't be there if the input is noise. > > >John > > I never heard the term "CW Stick". Cant get any reference online too. It > will be helpful if you could explain a bit more on this.Had to laugh at that one. It just means an unmodulated carrier in the spectrum. John
Reply by ●December 16, 20092009-12-16
>Demodulate the signal incoherently, compute RMS of error vector, compare>with a threshold. You might want separate thresholds for phase and >amplitude of error.What is the meaning of demodulating incoherently in baseband? I dont intend to use a training sequence. In that case how do I compute the error vector? In the case of pi/4 dqpsk, an angle between 0 and pi/4 can be either 0 or pi/4. Let me clarify my requirement in more detail. Mobile radio scans all the channels in a band to find out the available networks/sites. Currently I select the channels by calculating the Received Signal Strength of the channel which is not very reliable. Also the scanning has to be completed fast so I cannot afford to do a correlation to search for a training sequence for each channel in the band. So if we could identify the presence of a modulated signal in a simple way, the result can be used along with RSSI to take a good decision.
