amplitude modulation
given absolutely no noise on the line and perfect equipment on both ends, bit rate is possibly infinite, this never actually happens...
noisy channel - signal could be corrupted, need filters to remove noise
AWGN - additive white gaussian noise
gaussian distribution N(0,sigma) e^(-(N^2)/2sigma^2)/sqrt(n2pi)
checking for error again - intersection point is x = (A0+A1)/2
error only occurs when generated signal (or noisy signal) is greater than halfway between the two expected values
probability of error - use lookup tables to find G, integral of gaussian distribution
can use erfc in matlab to calculate it, complementary error function
input t = (x-A0)/(sqrt(2)sigma)
error probability = erfc((A1-A0)/(2sqrt(2)sigma))/2

dr = (A1-A0)/sigma -> POE = erfc(dr/2sqrt(2))/2

analog to digital conversion
sampling, quantization, encoding
errors caused by jamming signals into bins to correspond to quantized bits
when regenerating the signal from the quantized form, it will rarely be identical, this error is called quantization error
quantization errors tend to follow uniform distribution, all values of error are equally likely