|
From: | Bob McGwier |
Subject: | Re: [Discuss-gnuradio] Using DSP for precise zero crossing, measurement? |
Date: | Mon, 18 Sep 2006 13:44:09 -0400 |
User-agent: | Thunderbird 1.5 (X11/20051201) |
John,Achilleas implies how to compute the "score" for the parameters A1, A2, t1, and t2 and that the problem is to minimize the distance between r and s, but he does not say how to reestimate or compute the best set of parameters. That is all the meat to the problem (the minimization of this L^2 norm). Since you are the smartest lawyer with no engineering training I have ever known in so far as your engineering abilities are concerned you can probably do this but I suspect some further guidance will make this easier.
73's Bob (N4HY) John Ackermann N8UR wrote:
Achilleas Anastasopoulos wrote:John,If want to measure the time difference between two sine waves in noise and accuracy is your primary objective, then you should start with the"optimal" solution to the problem and not with an ad-hoc technique such as measuring the zero crossings. In the simplest scenario, if your model looks like: s(t) = s(t;A1,A2,t1,t2) = A1 sin(w (t-t1)) + A2 sin(w (t-t2)) r(t)=s(t)+n(t) and if you assume Gaussian noise n(t), then if your observation is t in (0,T) the (maximum likelihood) estimate should minimize the squared Euclidean distance int_0^T |r(t)-s(t)|^2 dt or in discrete time (with appropriate sampling Ts) sum_k |r(k Ts)-s(k Ts)|^2 There are several simplifications (eg, if you know apriori that the 2 amplitudes are the same, or if you know oneof t1 or t2, etc) you can try (depending on the required accuracy) but this is the safest way to approach the problem.AchilleasAchilleas, thanks for the answer, which it will take me a little while to digest as I'm not a mathematician -- more accurately, I'm a mathematical anti-prodigy! However, I see generally what you're saying and measuring zero crossings isn't the way it "has" to be done, it's the way it's been done in the past for this application.One factor is that we will *not* know that the two amplitudes are the same; while we can aim for that, in some cases they may be several dB apart because of the nature of the device under test and the reference signal. And, we don't want to add additional amplifiers, etc. into the system unless we really have to because they could contribute their own noise to the result.My posting a few minutes ago describes in more detail just what we're trying to accomplish. Hopefully that will help explain the goal.Thanks! John _______________________________________________ Discuss-gnuradio mailing list address@hidden http://lists.gnu.org/mailman/listinfo/discuss-gnuradio
-- Robert W. McGwier, Ph.D. Center for Communications Research 805 Bunn Drive Princeton, NJ 08540 (609)-924-4600 (sig required by employer)
[Prev in Thread] | Current Thread | [Next in Thread] |