I am detecting an input signal from a sensor and plan to perform real time peak detection on it. But for some reason the peak detection calculations induces a lot of noise in the input signal.
Do you have any solution to prevent this ??

Hey there,
I am trying to use your library implementation of the PeakDetection algorithm and I have to make some modifications for my needs. When trying to understand the code I stumbled upon your implementation of the getStd() method.

As far as I understand the algorithm this should just collect the sum of squared differences between stored values and the average value and then divide it by the amount of values to get the average squared difference of the values to the mean.
The square root of this is then returned as the average difference of the signal to the mean.

Other peoples code looks like this, and is easily understandable for me:
https://stackoverflow.com/questions/22583391/peak-signal-detection-in-realtime-timeseries-data/54507140#54507140 and I can understand it really well...

float stddev(float data[], int len) {
    float the_mean = mean(data, len);
    float standardDeviation = 0.0;

    int i;
    for(i=0; i<len; ++i) {
        standardDeviation += pow(data[i] - the_mean, 2);
    }
    return sqrt(standardDeviation/len);
}

Your implementation goes like this:

double PeakDetection::getStd(int start, int len) {
  double x1 = getAvg(start, len); 
  double x2 = getPoint(start, len);
  double powx1 = x1 * x1;
  double std = x2 - powx1;
  if (std > -EPSILON && std < EPSILON)
    return 0.0;
  else
    return sqrt(x2 - powx1);
}

and I really dont understand it. What is this getPoint() method doing?

double PeakDetection::getPoint(int start, int len) {
  double xi = 0.0;
  for (int i = 0; i < len; ++i)
    xi += data[(start + i) % lag] * data[(start + i) % lag];
  return xi / len;
}

If i understand correctly it returns the average squared values of all collected values. I don't get why this is necessary. Can you help me to understand this?

thank you for any help on this,
paul

__ EDIT:
Sorry, I just recalculated it with some test data and I am surprised your function gave the right result. Seems like a mathematical trick I would like to understand...
This is some test scenario with caluclations as I assume them:

testdata = [ 1, 7, 4, 2]
avg -> 14/4 = 3,5
stdDeviation -> -2.5, 3,5, 0.5, -1,5 -> 6.25 + 12,25 + 0,25 + 2,25 = 5,25
root(5,25) => 2,29128784747792

Your code results in :
x1 = 3,5
x2 -> 1 + 49 + 16 + 4 -> 17,5
powx1 = 12,25
std -> 17,5 - 12,25 = 5,25
root(5,25) => 2,29128784747792

const int DEFAULT_INFLUENCE = 0.5;
const int DEFAULT_EPSILON = 0.01;

How does that work?

I have a sinewave from mpu9250 sensor which I need to calculate the max of during live sensor readings. I'm trying to understand how lag threshold influence values work in your library and what i should set them to. is there any other sources? In example they are peakDetection.begin(48, 3, 0.6); and have no idea why.