M-FSK PHI(t) = Acos(2tpi*fi), i=<1,M> fi = f1 + (i-1)delf PSD looks sinish with pulse centers at each fi want orthogonal signals, for example: int<0,T>(Acos(2tpi*fm)Acos(2tpi*fn))dt = 0 means that they're orthogonal ends up being AA/2 * int<0,T>(cos(2tpi*(fn+fm)) + cos(2tpi*(fm-fn))) ~~ AA/2 * (sin(2Tpi*(fm-fn))/(2pi*(fm-fn))) = 0 need 2pi*(fm-fn)T, this means that if you space frequencies by 1/2T, the signals will be orthogonal M-PSK Si(t) = sqrt(2E/T)cos(wct + ((2i+1)/M)pi), i=0..M-1 = sqrt(2E/T)(cos(wct)cos((2i+1)pi/M) - sin(wct)sin((2i+1)pi/M)) cos piece of last eqn ^^ is the in phase component, the sin portion is the quadrature component 4-PSK is equivalent to 2 binary-ask sigs organized in a circle around the origin, polar looking graph Q-ASK aka QAM M = M'M" s(t) =Auicos(wct)+Avjsin(wct) ui in {-(M'-1), -(M'-3)..-1,1,3..(M'-1)} ditto for vi and M" usually arranged in a rectangle, further from the origin means it takes more energy to send it 8QAM ui in {-3,-1,1,3} vi in {-1,1} 16QAM ui in {-3,-1,1,3} vi in {-3,-1,1,3} oh shit...probability distribution functions watch for mudkips and lions mu is mean, sigma^2 is the variance PDF is prob density f(), probability that X lies between a and b if mean is 0, variance is equal to avg normalized power thermal noise caused by thermal motion of electrons in all dissipative components has following tendencies: mean is 0 has gaussian PDF has avg normalized PSD that is constant=No/2 from DC to ~10^12Hz called additive white gaussian noise filtering of random sigs if a random sig with mean 0 goes through a LTI filter, the output will also have mean 0, but it will no longer be white noise. it will still be gaussian gaussian function doesn't have a closed form integration we use a tabulation instead tabulation known as complementary error function aka Q function, values for mu=0, variance=1 Q(-a) = 1-Q(a) performance criterion transmitter input chosen from finite set of possible symbols receiver obj: accurately det which symbol was transmitted figure of merit: prob of error in decision, often interested in prob of bit error or bit error rate (BER) general receiver binary sig transmitted with thermal noise added to channel