Observations 1)We get a similar picture with any c(t that is periodic with period T' 2) As long as oc=2 /T>20M, there is no overlap in the shifted and scaled replicas of XGo. Consequently, assuming ao*0 x(t) 0/2 c(t)=2aejkot x(t can be recovered by passing y(t)through a LPF 3)Pulse Train Modulation is the basis for Time-Division Multiplexing Assign time slots instead of frequency slots to different channels, e. g. AT&t wireless phones 4)Really only need samples (x(nT))when oc>2 o*adulation Pulse amplitudeObservations 1) We get a similar picture with any c(t) that is periodic with period T 4) Really only need samples {x(nT)} when ωc > 2 ωM ⇒ Pulse Amplitude Modulation x(t) can be recovered by passing y(t) through a LPF 2) As long as ωc = 2π/T > 2ωM, there is no overlap in the shifted and scaled replicas of X(jω). Consequently, assuming ao ≠ 0: 3) Pulse Train Modulation is the basis for Time-Division Multiplexing — Assign time slots instead of frequency slots to different channels, e.g. AT&T wireless phones