# Trapezoidal Integral of Discrete Values

Hi guys, so I'm trying to create this thing that gets accelerometer values and integrate those values at about 25 samples a second. From what I understand if I integrate the values, that means each second I have 25 values I want to integrate. I read this research paper where someone used the trapezoidal integration to do it but I've been trying to read up on it and I guess I'm having some problems. Firstly can I use the trapezoidal integration to do it?

Secondly this is the equation they used:
y(n) = y(n-1) + (1/(2fs))*(x(x-1)+x(n)) , where n>0

so I guess fs is my sampling frequency which would be 25 and x is my integrand. I would assume n is my value so 1 or 2 or 3rd sample but I'm worried about y(n-1) and the x(n) part since the start of the equation is y(n).

Basically if say my values were
Fs=25
n=1
x= 32

then is it right to say
y(n) = y(1-1) + (1/(2*25))*(32(32-1)+32(1))
or is the y(n-1) similar to the y(n) part and I have to sort of work out the integration together?