Standard Analysis Plugins

The following standard analysis plugins are distributed with SAM.

Autocorrelation

This plugin plots the autocorrelation coefficients for the data against the time lag, where the time lag $k$ has a range from 0 to the final time at which data was dumped:

$$r_k = \frac{n}{n-k}\frac{\sum_{i=1}^{n-k} \ {(x_i - \overlin... ...i+k}- \ \overline{x})}}{\sum_{i=1}^n {(x_i - \overline{x})^2}}$$

where $r_k$ is the lag-$k$ autocorrelation coefficient and the $x_i$ are the $n$ data values (with the scaling factor of $\frac{n}{n-k}$ added in).

CDF (Cumulative Distribution Function)

This plugin calculates the Cumulative Distribution Function of the data:

$$cdf(x) = Pr[X \leq x]$$

where $cdf(x)$ is the probability that the variable takes a value less than or equal to the data value $x$, displaying it as a histogram, with the values of the data along the X-axis, and the probability along the Y-axis.

The only option for the CDF analysis is the number of bins, as shown below. It must be a positive integer.

Figure 7: Additional settings to set for CDF plugin

Spectral Analysis (FFT)

This plugin performs a FFT (fast fourier transform) on the data.

Linear Interpolation

This plugin performs linear interpolation upon the data, essentially increasing the number of data points to ensure a more uniform time interval between data points. It does this by first defining a value meanstep, which is the arithmetic mean of the five smallest $\Delta t = t_i - t_{i-1}$ values. Whenever two consecutive values $t_i$ and $t_{i-1}$ are separated by more than one and a half meansteps, an extra data point $x(t_{i-1}+ \mathtt{meanstep})$ is added to our set of values $x(t)$, such that $x(t_{i-1}+ \mathtt{meanstep}) = \mathtt{meanstep}*(x(t_i) - x(t_{i-1}))/(t_i - t_{i-1})$.

This is automatically used by the Autocorrelation and Spectral Analysis plugins.

Mean

This plugin calculates the mean of the data as follows:

$$\overline{x} = \frac{1}{n}\sum_{i=1}^n x_i$$

where $\overline{x}$ is the mean and the $x_i$ are the $n$ data values. A single value is calculated for each stream.

PDF (Probability Distribution Function)

This function calculates the Probability Distribution Function for the data:

$$pdf(x) = Pr[X = x]$$

where $pdf(x)$ is the probability that the variate takes the data value $x$, displaying it as a histogram, with the values of the data along the X-axis, and the probability along the Y-axis.

It has the settings window below, allowing you to set the number of bins used for the calculation:

Figure 8: Additional settings to set for PDF plugin

Note that the number of bins must be a positive integer.

No Analysis

The `No Analysis' analysis plugin is an identity plugin, i.e. no actual analysis is performed on the data. This plugin is useful to use if, for example, you just want to output the window size vs time for a TCP stream.

Variance

This plugin calculates the variance of the data as follows:

$${\rm variance} = \frac{1}{n-1}\sum_{i=1}^n {(x_i - \overline{x})^2}$$

where $\overline{x}$ is the mean as described by the Mean plugin above, and the $x_i$ are the $n$ data values. A single value is calculated for each stream.