Statistical analysis of discrete signals

hi everyone,

            i am working on a project where i have to acquire signals from the sensors and do some statistical analysis. i want to know how can we find the amplitude distribution of a discrete signal.

Answer 1

[Walter Roberson]

Sounds like a job for hist() against the original (amplitude) signal.

[phani]

is there any tool available in matlab to find the amplitude distribution

[Jan]

Well, did anybody mention HIST or HISTC already?

[Walter Roberson]

Phani, there is no tool (that I know of) in MATLAB to find amplitude distribution specifically. There are, however, general tools to find distributions, and you can apply those general tools to your amplitude data. Presuming that is, that you are reading in amplitude data and not something else.

Perhaps the question we should be asking is what _you_ mean by "amplitude distribution". Are you referring to the signal measurements recorded by the sensors, or are you referring to the "original" amplitude before it was converted to a sensor reading? For example are you looking to examine the internal distance moved by the disc inside a microphone, reflecting the sound pressure level? Or perhaps you want statistics on the dB of sound emitted by a speaker as opposed to the measurement of the +/- 0.3V line-in signal ?

[phani]

walter, thanks for your information. i am trying to know the amplitude distribution of the sensor signal. the sensor will converts the change in the physical parameters to volts signal. i want to find the amplitude distribution of that volts signal.

[Walter Roberson]

The A/D convertor will compare the voltage to a pre-selected voltage range (adjustable in some convertors, fixed in others) and will scale the results to an integer value, usually 8 or 10 or 12 bits. It is those integer values that are what your program receives, and those values are proportional to the voltage (but possibly have been shifted so that negative and positive voltages could both be represented.) Do you need to calculate back to the voltages, or is it sufficient to work with the scaled (and possibly shifted) proportional values?

If you can work with the scaled and shifted values (which will very likely be the case), then you can use hist() or histc() on the input values.

Have you read the documentation for hist() and histc() ?