Hi guys I'm trying to find a solution for detecting a sinusoidial signal under non-gaussian noise, for example sinusoidial noise, consider the following model: y1(t)=A*exp(j*w*t)+B*exp(j*w*t)+n(t) y2(t)=B*exp(j*w*t)+n(t) I want to detect if y1 or y2 is present, do you think that ML detection is applicable here (p(r/y1)/p(r/y2)><Th)? how do I calculate the ML solution here? Thanks, Kal
maximum likelihood sinusoidal signal detection with sinusoidal noise
Started by ●September 26, 2011
Reply by ●September 26, 20112011-09-26
kalore wrote:> Hi guys > I'm trying to find a solution for detecting a sinusoidial signal under > non-gaussian noise, for example sinusoidial noise, consider the following > model: > y1(t)=A*exp(j*w*t)+B*exp(j*w*t)+n(t) > y2(t)=B*exp(j*w*t)+n(t) > I want to detect if y1 or y2 is present, do you think that ML detection is > applicable here (p(r/y1)/p(r/y2)><Th)? > how do I calculate the ML solution here? > Thanks, Kal > > > >
Reply by ●September 26, 20112011-09-26
On Mon, 26 Sep 2011 09:56:26 -0500, kalore wrote:> Hi guys > I'm trying to find a solution for detecting a sinusoidial signal under > non-gaussian noise, for example sinusoidial noise, consider the > following model: > y1(t)=A*exp(j*w*t)+B*exp(j*w*t)+n(t) > y2(t)=B*exp(j*w*t)+n(t) > I want to detect if y1 or y2 is present, do you think that ML detection > is applicable here (p(r/y1)/p(r/y2)><Th)? how do I calculate the ML > solution here? Thanks, KalI don't know. "A solution" to your detection problem is s = 0 -- but that's not what you want. Maximum likelihood works if you know something of the nature of the noise. Yes, maximum likelihood _may_ work, but whether it'll be good enough depends on a whole world of stuff that you haven't told us. -- www.wescottdesign.com
Reply by ●September 27, 20112011-09-27
Reply by ●September 27, 20112011-09-27
> > >kalore wrote: > >> Hi guys >> I'm trying to find a solution for detecting a sinusoidial signal under >> non-gaussian noise, for example sinusoidial noise, consider thefollowing>> model: >> y1(t)=A*exp(j*w*t)+B*exp(j*w*t)+n(t) >> y2(t)=B*exp(j*w*t)+n(t) >> I want to detect if y1 or y2 is present, do you think that ML detectionis>> applicable here (p(r/y1)/p(r/y2)><Th)? >> how do I calculate the ML solution here? >> Thanks, Kal >> >> >> >>I think you can do it, but it is not required. . Follow the standard ML derivation.Y1 will have the distribution at A+B and y2 will have at B. You can derive from here... Regards Hozefa>