DIGITAL COMMUNICATION Block Diagram Functional model of pass-band data transmission system. 2 Digital Modulation Technique : -ASK : Amplitude shift keying . -FSK : Frequency shift keying . -PSK : Phase shift keying . -QAM: Quadrature amplitude modulation. (Special case of Hybrid Modulation Techniques) 3 What do we want to study? We are going to study and compare different modulation techniques in terms of Power Spectrum. Bandwidth efficiency: Ro B Bits/s/Hz Probability of errors. 4 Complex Representation of bandpass Signals: Let 𝑔(𝑡) be any signal. Define 𝑔+ 𝑡 = 𝑔 𝑡 + 𝑗 ∙ 𝑔(𝑡) which called the analytic signal (pre-envelope) of 𝑔(𝑡). 𝑔(𝑡) is the Helbert transform of 𝑔(𝑡). 𝑔 𝑡 = 1 𝜋𝑡 ⊗ 𝑔(𝑡), ℱ 𝑔(𝑡) = ℱ 𝐺 𝑓 =ℱ 𝑗 1 ℱ 𝜋𝑡 1 𝜋𝑡 1 𝜋𝑡 ⊗convolution. ∙ ℱ 𝑔(𝑡) ∙ 𝐺(𝑓) 𝑠𝑔𝑛 𝑓 1 ℱ 𝜋𝑡 − 𝑗 ∙ 𝑠𝑔𝑛 𝑓 5 Cont. ℱ 1 𝜋𝑡 −𝑗, 𝑓 > 0 = 0, 𝑓 = 0 𝑗, 𝑓 < 0 sgn(f) +1 f -1 𝐺 𝑓 = −𝑗 ∙ 𝑠𝑔𝑛 𝑓 ∙ 𝐺 𝑓 −𝑗𝐺 𝑓 = 𝐺 𝑓 𝑒 −𝑗𝜋 2 , 𝑓 > 0 0, 𝑓 = 0 = 𝑗𝐺 𝑓 = 𝐺 𝑓 𝑒 +𝑗𝜋 2 , 𝑓 < 0 6 Cont. ℱ 𝑔+ (𝑡) = 𝐺 𝑓 + 𝑠𝑔𝑛 𝑓 ∙ 𝐺(𝑓) 2𝐺 𝑓 , 𝑓 > 0 𝐺+ 𝑓 = 𝐺 0 , 𝑓 = 0 0, 𝑓 < 0 Also define 𝑔− 𝑡 = 𝑔 𝑡 − 𝑗 ∙ 𝑔(𝑡) called the –ve pre-envelope. 𝐺− 𝑓 = 𝐺 𝑓 − 𝑠𝑔𝑛 𝑓 ∙ 𝐺(𝑓) 0, 𝑓 > 0 𝐺− 𝑓 = 𝐺 0 , 𝑓 = 0 2𝐺 𝑓 , 𝑓 < 0 𝑔− 𝑡 = 𝑔+ ∗ (𝑡) 7 Cont. Let 𝑔(𝑡) be a bandpass signal. 𝐺(𝑓) 𝐺(𝑓𝑐 ) −𝑓𝑐 − 𝑓𝑜 −𝑓𝑐 𝑓 −𝑓𝑐 + 𝑓𝑜 𝑓𝑐 − 𝑓𝑜 2 𝐺(𝑓𝑐 ) 𝑓𝑐 𝑓𝑐 + 𝑓𝑜 𝐺+ (𝑓) 𝑓 𝑓𝑐 − 𝑓𝑜 2 𝐺(𝑓𝑐 ) −𝑓𝑜 𝑓𝑐 𝑓𝑐 + 𝑓𝑜 𝐺 (𝑓) 𝑓𝑜 𝑓 𝑓𝑐 − 𝑓𝑜 𝑓𝑐 𝑓𝑐 + 𝑓𝑜 8 Cont. Let 𝐺+ 𝑓 + 𝑓𝑐 = 𝐺 𝑓 = ℱ 𝑔(𝑡) 𝑔(𝑡) is the equivalent low-pass signal of 𝑔(𝑡) called the complex envelope of 𝑔(𝑡). 𝑔 𝑡 = ℱ −1 𝐺+ (𝑓 + 𝑓𝑐 ) = 𝑔+ (𝑡)𝑒 −𝑗2𝜋𝑓𝑐𝑡 𝑔 𝑡 𝑒 𝑗2𝜋𝑓𝑐 𝑡 = 𝑔+ 𝑡 = 𝑔 𝑡 + 𝑗 𝑔(𝑡) 𝑔 𝑡 = 𝑅𝑒 𝑔+ 𝑡 = 𝑅𝑒 𝑔 𝑡 𝑒 𝑗2𝜋𝑓𝑐 𝑡 9 Cont. 𝑔 𝑡 in general is complex-valued function. ⇒ 𝑔 𝑡 = 𝑔𝐼 𝑡 + 𝑗 𝑔𝑄 (𝑡) 𝑔𝐼 𝑡 : is real lowpass signal called the In-phase component of 𝑔(𝑡). 𝑔𝑄 𝑡 : is real lowpass signal called the Quadrature component of 𝑔(𝑡). ⇒ 𝑔 𝑡 = 𝑅𝑒 𝑔𝐼 𝑡 + 𝑗 𝑔𝑄 (𝑡) 𝑒 𝑗2𝜋𝑓𝑐 𝑡 = 𝑅𝑒 𝑔𝐼 𝑡 + 𝑗 𝑔𝑄 (𝑡) cos 2𝜋𝑓𝑐 𝑡 + 𝑗 sin(2𝜋𝑓𝑐 𝑡) ⇒ 𝑔 𝑡 = 𝑔𝐼 (𝑡) cos 2𝜋𝑓𝑐 𝑡 − 𝑔𝑄 (𝑡) sin(2𝜋𝑓𝑐 𝑡) ⇒ 𝑔 𝑡 = 𝐼𝑚 𝑔+ 𝑡 = 𝐼𝑚 𝑔 𝑡 𝑒 𝑗2𝜋𝑓𝑐 𝑡 ⇒ 𝑔 𝑡 = 𝑔𝐼 (𝑡) sin 2𝜋𝑓𝑐 𝑡 + 𝑔𝑄 (𝑡) cos(2𝜋𝑓𝑐 𝑡) 10 Cont. To get 𝑔𝐼 𝑡 and 𝑔𝑄 (𝑡) from 𝑔(𝑡) use the following system. LPF 𝑔𝐼 (𝑡) LPF 𝑔𝑄 (𝑡) 2 cos 𝜔𝑐 𝑡 𝑔(𝑡) ~ π/2 −2 sin 𝜔𝑐 𝑡 11 Cont. To get back 𝑔(𝑡) from 𝑔𝐼 𝑡 and 𝑔𝑄 (𝑡) use the following system. 𝑔𝐼 (𝑡) cos 𝜔𝑐 𝑡 + ~ ∑ -π/2 𝑔𝑄 (𝑡) 𝑔(𝑡) - sin 𝜔𝑐 𝑡 12 Cont. Since 𝑔 𝑡 = 𝑔𝐼 𝑡 + 𝑗 𝑔𝑄 (𝑡) is complex it can be written in the polar form as: 𝑔 𝑡 = a(t)𝑒 𝑗𝜙(𝑡) Where 𝑎 𝑡 = 𝑔2 𝐼 𝑡 + 𝑔2 𝑄 𝑡 Called Natural Envelope of 𝑔(𝑡). 𝜙 𝑡 = 𝑔𝑄 (𝑡) −1 tan 𝑔𝐼 𝑡 Called the phase of 𝑔(𝑡) ⇒ 𝑔 𝑡 = 𝑎 𝑡 cos 2𝜋𝑓𝑐 𝑡 + 𝜙(𝑡) 𝑔𝐼 𝑡 = 𝑎 𝑡 cos 𝜙 𝑡 and 𝑔𝑄 𝑡 = 𝑎 𝑡 sin 𝜙 𝑡 13 Cont. ∞ 𝐺 𝑓 = −∞ 𝑔(𝑡)𝑒 −𝑗2𝜋𝑓𝑡 𝑑𝑡 ∞ = −∞ 𝑅𝑒 𝑔(𝑡)𝑒 𝑗2𝜋𝑓𝑐 𝑡 𝑒 −𝑗2𝜋𝑓𝑡 𝑑𝑡 1 But 𝑅𝑒 𝑥 = 𝑥 + 𝑥 ∗ 2 1 ∞ ⇒ 𝐺 𝑓 = −∞ 𝑔 𝑡 𝑒 𝑗2𝜋𝑓𝑐𝑡 + 𝑔∗ (𝑡)𝑒 −𝑗2𝜋𝑓𝑐 𝑡 2 1 ⇒ 𝐺 𝑓 = 𝐺 𝑓 − 𝑓𝑐 + 𝐺 ∗ (−𝑓 − 𝑓𝑐 ) 2 𝑒 −𝑗2𝜋𝑓𝑡 𝑑𝑡 14 Power Spectral Density Define PSD: 𝑆𝑔 𝑓 ≜ 𝐺 𝑓 𝑆𝑔 𝑓 = 1 2 2 = 𝐺(𝑓)𝐺 ∗ (𝑓) 𝐺 𝑓 − 𝑓𝑐 + 𝐺 ∗ (−𝑓 − 𝑓𝑐 ) 1 2 𝐺 ∗ 𝑓 − 𝑓𝑐 + 𝐺(−𝑓 − 𝑓𝑐 ) Since 𝐺 𝑓 − 𝑓𝑐 and 𝐺 ∗ (−𝑓 − 𝑓𝑐 ) do not overlap ⇒ 𝑆𝑔 𝑓 = ⇒ 𝑆𝑔 𝑓 = 1 1 ∗ ∗ 𝐺 𝑓 − 𝑓𝑐 𝐺 (𝑓 − 𝑓𝑐 ) + 𝐺 (−𝑓 4 4 2 2 1 1 𝐺 𝑓 − 𝑓𝑐 + 𝐺 −𝑓 − 𝑓𝑐 4 4 1 4 − 𝑓𝑐 )𝐺 −𝑓 − 𝑓𝑐 1 4 ⇒ 𝑆𝑔 𝑓 = 𝑆𝑔 𝑓 − 𝑓𝑐 + 𝑆𝑔 −𝑓 − 𝑓𝑐 15 Representation of linear bandpass system Let ℎ(𝑡) be the impulse response of a linear system (filter) ⇒ ℱ ℎ(𝑡) = 𝐻 𝑓 its transfer function. Since ℎ(𝑡) is real ⇒ 𝐻 ∗ −𝑓 = 𝐻 𝑓 . 𝐻 𝑓 ,𝑓 > 0 Define 𝐻 𝑓 − 𝑓𝑐 = 0, 𝑓 < 0 ⇒ 𝐻∗ 0, 𝑓>0 −𝑓 − 𝑓𝑐 = 𝐻 ∗ −𝑓 , 𝑓 < 0 0, 𝑓>0 = 𝐻 𝑓 ,𝑓 < 0 So 𝐻 𝑓 = 𝐻 𝑓 − 𝑓𝑐 + 𝐻 ∗ −𝑓 − 𝑓𝑐 ⇒ ℎ 𝑡 = ℎ 𝑡 𝑒 𝑗2𝜋𝑓𝑐𝑡 + ℎ∗ 𝑡 𝑒 −𝑗2𝜋𝑓𝑐𝑡 = 2𝑅𝑒 ℎ(𝑡)𝑒 𝑗2𝜋𝑓𝑐𝑡 16 Response of a bandpass system to a bandpass signal 𝑔(𝑡) is a bandpass signal, 𝑔(𝑡) is its equivalent lowpass signal. ℎ 𝑡 is an impulse response of bandpass system, ℎ(𝑡) is its equivalent lowpass impulse response. 𝑟(𝑡) the output of a bandpass system is also bandpass. ⇒ 𝑟 𝑡 = 𝑅𝑒 𝑟(𝑡)𝑒 𝑗2𝜋𝑓𝑐 𝑡 𝑟 𝑡 =𝑔 𝑡 ⊗ℎ 𝑡 = ∞ 𝑔 −∞ 𝜏 ℎ 𝑡 − 𝜏 𝑑𝜏 𝑅 𝑓 = ℱ 𝑟(𝑡) = 𝐺 𝑓 𝐻 𝑓 ⇒𝑅 𝑓 = 1 2 𝐺 𝑓 − 𝑓𝑐 + 𝐺 ∗ (−𝑓 − 𝑓𝑐 ) 𝐻 𝑓 − 𝑓𝑐 + 𝐻∗ −𝑓 − 𝑓𝑐 𝑔(𝑡) 𝑟(𝑡) ℎ(𝑡) 17 Cont. But for narrowband signal and system: 𝐺 𝑓 − 𝑓𝑐 𝐻 ∗ −𝑓 − 𝑓𝑐 = 0 And 𝐺 ∗ −𝑓 − 𝑓𝑐 𝐻 𝑓 − 𝑓𝑐 = 0 ⇒𝑅 𝑓 = 1 2 𝐺 𝑓 − 𝑓𝑐 𝐻 𝑓 − 𝑓𝑐 + 𝐺 ∗ (−𝑓 − 𝑓𝑐 )𝐻 ∗ (−𝑓 − 𝑔(𝑡) 𝑟(𝑡) ℎ(𝑡) 18 Representation of Bandpass Stationary Random Process Let 𝑛(𝑡) be a sample function of a wide-sense stationary random process with zero mean and PSD of 𝜓𝑛 𝑓 . The PSD is concentrated about ±𝑓𝑐 and zero outside a certain band. ⇒ 𝑛 𝑡 = 𝑎 𝑡 cos 2𝜋𝑓𝑐 𝑡 + 𝜃(𝑡) = 𝑥 𝑡 cos 2𝜋𝑓𝑐 𝑡 − 𝑦 𝑡 sin 2𝜋𝑓𝑐 𝑡 = 𝑅𝑒 𝑛(𝑡)𝑒 𝑗2𝜋𝑓𝑐𝑡 𝑛 𝑡 = 𝑥 𝑡 + 𝑗𝑦(𝑡) 𝑛 𝑡 is the complex envelope (lowpass) 𝑥 𝑡 is the In phase component (lowpass) 𝑦(𝑡) is the Quadrature component (lowpass) 19 Cont. 𝑅𝑛𝑛 𝜏 = 𝐸 𝑛 𝑡 𝑛(𝑡 − 𝜏) = 𝐸 𝑥 𝑡 cos 2𝜋𝑓𝑐 𝑡 − 𝑦 𝑡 sin 2𝜋𝑓𝑐 𝑡 𝑥 𝑡 − 𝜏 cos 2𝜋𝑓𝑐 (𝑡 − 𝜏) − 𝑦 𝑡 − 𝜏 sin(2𝜋𝑓𝑐 (𝑡 − 20 Cont. 𝑅𝑛𝑛 𝜏 ⇒ 𝑅𝑛𝑛 𝑅𝑥𝑥 𝜏 𝑅𝑦𝑦 𝜏 = 𝐸 𝑛 𝑡 𝑛(𝑡 − 𝜏) −𝜏 = 𝐸 𝑛 𝑡 𝑛(𝑡 + 𝜏) = 𝑅𝑛𝑛 𝜏 even function = 𝐸 𝑥 𝑡 𝑥(𝑡 − 𝜏) even function = 𝐸 𝑦 𝑡 𝑦(𝑡 − 𝜏) even function But: 𝑅𝑦𝑥 −𝜏 = 𝐸 𝑦 𝑡 𝑥(𝑡 + 𝜏) = 𝐸 𝑥 𝛾 𝑦(𝛾 − 𝜏) = 𝑅𝑥𝑦 𝜏 = −𝑅𝑦𝑥 𝜏 ⇒odd function And 𝑅𝑥𝑦 (𝜏) is also odd. 21 Cont. 𝑛 𝑡 = 𝑥 𝑡 + 𝑗𝑦(𝑡) 𝑅𝑛𝑛 𝜏 = = = 1 𝐸 𝑛(𝑡)𝑛∗ (𝑡 − 𝜏) 2 1 𝐸 𝑥 𝑡 + 𝑗𝑦(𝑡) 2 1 𝑅𝑥𝑥 𝜏 + 𝑅𝑦𝑦 𝜏 2 𝑥 𝑡 − 𝜏 − 𝑗𝑦(𝑡 − 𝜏) − 𝑗𝑅𝑥𝑦 𝜏 + 𝑗𝑅𝑦𝑥 (𝜏) But 𝑅𝑥𝑥 𝜏 = 𝑅𝑦𝑦 𝜏 , 𝑅𝑦𝑥 𝜏 = −𝑅𝑥𝑦 𝜏 ⇒ 𝑅𝑛𝑛 𝜏 = 𝑅𝑥𝑥 𝜏 + 𝑗𝑅𝑦𝑥 (𝜏) 𝑅𝑛𝑛 𝜏 = 𝑅𝑥𝑥 𝜏 cos(2𝜋𝑓𝑐 𝜏) − 𝑅𝑦𝑥 𝜏 sin 2𝜋𝑓𝑐 𝜏 ⇒ 𝑅𝑛𝑛 𝜏 = 𝑅𝑒 𝑅𝑛𝑛 𝜏 𝑒 𝑗2𝜋𝑓𝑐𝜏 22 Cont. 𝑅𝑛𝑛 𝜏 = 𝑅𝑛∗ 𝑛 (−𝜏) since 𝑅𝑦𝑥 (𝜏) is odd function. PSD of 𝑛(𝑡): 𝑆𝑛 𝑓 = ℱ 𝑅𝑛𝑛 𝜏 = ∞ 𝑅 𝜏 −∞ 𝑛𝑛 −𝑗2𝜋𝑓𝜏 ∞ ∗ 𝑅 (−𝜏)𝑒 −∞ 𝑛𝑛 ∞ = −∞ 𝑅𝑛𝑛 −𝜏 𝑒 𝑗2𝜋𝑓𝜏 = 𝑆𝑛∗ 𝑓 = 𝑒 −𝑗2𝜋𝑓𝜏 𝑑𝜏 𝑑𝜏 𝑑𝜏 ∗ ⇒ 𝑆𝑛 𝑓 is real-valued function. PSD of 𝑛 𝑡 : 𝑆𝑛 𝑓 = ℱ 𝑅𝑛𝑛 𝜏 =ℱ 1 = 2 1 = 2 1 2 = ℱ 𝑅𝑒 𝑅𝑛𝑛 𝜏 𝑒 𝑗2𝜋𝑓𝑐𝜏 𝑅𝑛𝑛 𝜏 𝑒 𝑗2𝜋𝑓𝑐𝜏 + 𝑅𝑛∗ 𝑛 (𝜏)𝑒 −𝑗2𝜋𝑓𝑐𝜏 𝑆𝑛 𝑓 − 𝑓𝑐 + 𝑆𝑛∗ −𝑓 − 𝑓𝑐 𝑆𝑛 𝑓 − 𝑓𝑐 + 𝑆𝑛 −𝑓 − 𝑓𝑐 23 Power Spectral Density of Digital Carrier Systems: Φ(𝑡) = 𝑚(𝑡) cos(𝜔𝑐 𝑡) Where: Φ(𝑡): Modulated signal. 𝑚(𝑡): baseband message signal (Modulating signal). cos(𝜔𝑐 𝑡): un modulated carrier. 𝜔𝑐 : carrier frequency. The PSD is: SΦ(ω) Ψ𝑇(𝜔) 2 = lim 𝑇 𝑇→∞ 24 PSD of Digital Carrier Systems (Cont.) Where: Ψ𝑇 ω = ℱ ΦT(t) 𝑡 𝑇 ΦT(t) = Φ(t) ∙ 𝑟𝑒𝑐𝑡( ) 𝑡 𝑇 ΦT(t) = m(t) ∙ cos(𝜔𝑐 t) rect( ) ΦT(t) 𝑡 𝑟𝑒𝑐𝑡(𝑇) Φ(t) t -T/2 T/2 t -T/2 T/2 25 𝑡 ΦT(t) = m(t) rect( ) cos(𝜔𝑐 t) 𝑇 𝑚 𝑇 (𝑡) ℱ 𝑚 𝑇 (𝑡) = 𝑀𝑇 (𝜔) 1 Ψ𝑇(ω) = [𝑀𝑇 (ω − 𝜔𝑐 ) + 𝑀𝑇 (ω + 𝜔𝑐 )] 2 1 𝑀𝑇 (ω − 𝜔𝑐 ) + 𝑀𝑇 (ω + 𝜔𝑐 ) SΦ(ω) = lim 𝑇→∞ 4 T 2 𝑀 𝜔 : Band limited −𝜔𝐵 𝜔𝐵 𝜔 26 1 MT(ω−𝜔𝑐 ) SΦ(ω) = lim 𝑇→∞ 4 T If 𝜔𝑐 ≥𝜔𝐵 2 MT(ω+𝜔𝑐 ) + T 2 1 SΦ(ω) = [SM(ω−𝜔𝑐 ) + SM(ω+𝜔𝑐 )] 4 𝑤𝑜 2𝜋 1 Bandwidth = Hz Bandwidth = fo Hz −𝜔𝑜 PSD of m(t) 𝜔𝑜 𝜔 PSD of Φ(t) PΦ = 1 2 1/4 𝑃𝑚 −𝜔𝑐 − 𝜔𝑜 −𝜔𝑐 𝑤 Bandwidth = 2 (2𝜋𝑜) Hz Bandwidth = 2 fo Hz −𝜔𝑐 + 𝜔𝑜 𝜔𝑐 − 𝜔𝑜 𝜔𝑐 𝜔 𝜔𝑐 + 𝜔𝑜 27 Binary Digital Modulation Binary ASK: Φ𝐴𝑆𝐾 𝑡 = 𝑦 𝑡 + 𝐷𝐶 cos 𝜔𝑐 𝑡 𝑚(𝑡) = 𝑦 𝑡 cos 𝜔𝑐 𝑡 +(DC) cos 𝜔𝑐 𝑡 Where 𝑦(𝑡) is polar NRZ. 28 Φ𝐴𝑆𝐾 (𝑡) A -A A+DC Φ𝐴𝑆𝐾 (𝑡) DC -DC AM signal with m<1 2A Φ𝑂𝑂𝐾 (𝑡) -2A AM signal with m=1 On-Off keying m: modulation index 29 𝑃(𝜔) 𝑆𝑌 𝜔 = 𝑇 2 𝑃 𝜔 = 𝑇0 2 ∞ ℛ𝑛 𝑒 −𝑗𝑛𝜔𝑇 𝑛=−∞ For polar line code 𝑃 𝜔 = ℱ 𝑝 𝑡 ] could be any shape Let 𝑝 𝑡 = 𝐴 ∙ 𝑡−𝑇0 2 𝑟𝑒𝑐𝑡( ) 𝑇0 (NRZ rect) 𝑝(𝑡) 𝑃 𝜔 𝜔𝑇0 −𝑗𝜔𝑇0 2 = 𝐴 ∙ 𝑇0 𝑠𝑖𝑛𝑐( )𝑒 2 𝜔𝑇0 2 2 2 2 𝑃 𝜔 = 𝐴 ∙ 𝑇0 𝑠𝑖𝑛𝑐 ( ) 2 𝑆𝑌 𝜔 = 𝐴2 ∙ 𝑇0 𝑠𝑖𝑛𝑐 2 ( 𝑇0 2 t 𝑇0 𝜔𝑇0 ) 2 30 𝑆Φ (𝜔) −𝜔𝑐 − 2𝜋𝑅𝑜 −𝜔𝑐 −𝜔𝑐 + 2𝜋𝑅𝑜 (𝐷𝐶)2 𝜋 2 𝜔𝑐 − 2𝜋𝑅𝑜 𝜔𝑐 𝐴2 𝑇𝑜 4 𝜔𝑐 + 2𝜋𝑅𝑜 f 𝑓𝑐 − 𝑅𝑜 𝑓𝑐 𝑓𝑐 + 𝑅𝑜 EB = 2Ro 31 OOK: ΦOOK(t) = y(t) cos(𝜔𝑐 t) Here, y(t) is Unipolar NRZ. 𝑃(𝜔) SY(ω) = 4𝑇𝑜 ∞ 2 2𝜋 1+ 𝛿(𝜔 − 2𝜋𝑛𝑅𝑜 )] 𝑇𝑜 𝑛=−∞ 𝑡 − 𝑇0 2 𝑝 𝑡 = 2𝐴 ∙ 𝑟𝑒𝑐𝑡( ) 𝑇0 ∞ 𝜔𝑇𝑜 2𝜋 2𝜋𝑛 2 2 SY(ω) =𝐴 𝑇𝑜 𝑠𝑖𝑛𝑐 ( ) 1+ 𝛿(𝜔 − )] 2 𝑇𝑜 𝑇𝑜 𝑛=−∞ 𝐴2 𝜋 2 𝑆Φ (𝜔) 𝐴2 𝑇𝑜 4 −𝜔𝑐 − 2𝜋𝑅𝑜 −𝜔𝑐 −𝜔𝑐 + 2𝜋𝑅𝑜 𝜔𝑇 Since sinc( 2 𝑜) is zero at (2𝜋nR𝑜) then, one discrete component at 𝜔𝑐 𝜔𝑐 − 2𝜋𝑅𝑜 𝜔𝑐 𝜔𝑐 + 2𝜋𝑅𝑜 32 Binary ASK:Demodulation 1. 2. Coherent (Synchronous). Non-Coherent (Asynchronous) Using Envelope Detector (ED). Coherent: r(t) ΦASK(t) LPF 𝟐𝒄𝒐𝒔(𝝎𝒄 𝒕) ΦASK(t) = m(t) cos(𝜔𝑐 t) r(t) = m(t) cos(𝜔𝑐 t) ((2)cos(𝜔𝑐 t)) 1 r(t) = 2m(t)[ (cos(2𝜔𝑐 ) + cos(0))] = m(t) cos(2𝜔𝑐 t) + m(t) 2 And after the LPF: l(t) = m(t) = y(t) + DC l(t) y(t) + DC With noise it will replaced by a Match Filter (will explained later) 33 Non-coherant : ASK ENVELOPE DETECTOR yt DC 34 Constellation of Binary ASK 𝑠2 𝑠1 𝜓1 (𝑡) 35 Binary FSK 𝜑𝐹𝑆𝐾 = 𝐴 cos 𝜔𝑐 𝑡 + 𝑡 𝑘𝑓 −∞ 𝑦 𝛼 𝑑𝛼 ⇒ 𝐹𝑀 Where y(t) is polar NRZ. 36 1 0 0 1 1 0 1 0 0 1 1 0 polar A NRZ -A A 𝜑𝐵𝐹𝑆𝐾 -A A cos(𝜔1 𝑡) 𝜑𝑂𝑂𝐾1 𝑡 -A A 0 1 1 0 0 𝜑𝑂𝑂𝐾2 𝑡 -A cos(𝜔2 𝑡) 1 𝜑𝐵𝐹𝑆𝐾 𝑡 = 𝜑𝑂𝑂𝐾1 𝑡 + 𝜑𝑂𝑂𝐾2 (𝑡) Binary FSK(Cont.) "1": 𝐴 𝑝 𝑡 cos 𝜔1 𝑡 , 𝜆𝑇𝑜 < 𝑡 < (𝜆 + 1)𝑇0 "0": 𝐴 𝑝 𝑡 cos 𝜔2 𝑡 , 𝜆𝑇𝑜 < 𝑡 < 𝜆 + 1 𝑇𝑜 𝜔1 = 𝜔𝑐 + ∆𝜔 → mark frequincy. 𝜔2 = 𝜔𝑐 − ∆𝜔 → space frequincy. ∆𝜔 2𝜋 = ∆𝑓 Using Carson's Rule 𝐵𝐹𝑀 = 2∆𝑓 + 2𝐵 , B : Bandwidth of the base band signal Binary FSK(Cont.) Since FSK is an FM modulation with the base band signal y(t) which is polar NRZ → Bandwidth of y t is 𝐵 = 𝑅𝑜 𝐵𝑊 𝑜𝑓 𝐹𝑆𝐾 = 2∆𝑓 + 2𝑅𝑜 PSD of FSK is : PSD of 𝜑𝑂𝑂𝐾1 𝑡 + PSD of 𝜑𝑂𝑂𝐾2 𝑡 +𝑆12 𝜔 + 𝑆21 (𝜔) Cross PSDs 𝑆𝐵𝐹𝑆𝐾 𝜔 2𝑅0 2𝑅0 𝜔𝑐 − ∆𝜔 𝜔2 + 2π𝑅𝑜 𝜔𝑐 𝜔𝑐 + ∆𝜔 𝜔1 + 2π𝑅𝑜 𝜔 𝑓 𝑓2 𝑓2 + 𝑅𝑜 𝑓𝑐 𝑓1 − 𝑅𝑜 2∆𝑓 + 2𝑅𝑜 𝑓1 𝑓1 + 𝑅𝑜 Binary FSK(Cont.) With choosing Δω , ωc properly we can get rid of the discrete components at (ωc+ Δω) & (ωc-Δω) → (𝜔𝑐 +∆𝜔) − (𝜔𝑐 − ∆𝜔) → 2∆𝜔 = 𝑛(2𝜋𝑅𝑜 ) or 2∆𝑓 = 𝑛𝑅𝑜 → (Multiple of 𝑅𝑜 ) → ∆𝑓 = 𝑛𝑅𝑜 2 (No discrete component at (ωc+ Δω) & (ωc-Δω)) Δf increase → BW increase Δf decrease → BW decrease NonCoherent BFSK 𝑠1 𝑡 = 𝐴 cos 𝜔1 𝑡 + 𝜃1 , 𝜆𝑇𝑜 ≤ 𝑡 ≤ 𝜆 + 1 𝑇𝑜 , for “1” 𝑠2 𝑡 = 𝐴 cos 𝜔2 𝑡 + 𝜃2 , 𝜆𝑇𝑜 ≤ 𝑡 ≤ 𝜆 + 1 𝑇𝑜 , for “0” 𝜃1 , 𝜃2 are initial phases at 𝑡 = 0. These two signals are not coherent since 𝜃1 and 𝜃2 are not the same in general. The waveform is not continuous at bit transitions. This form of FSK is therefore called noncoherent or discontinuous FSK. It can be noncoherently demodulated. cos 𝜔1 𝑡 + 𝜃1 ~ 1 𝜑𝐵𝐹𝑆𝐾 (𝑡) 2 𝑐𝑜𝑠 𝜔2 𝑡 + 𝜃2 ~ 𝐼𝑘 43 Coherent BFSK The two signals have the same initial phase 𝜃 at t = 0. 𝑠1 𝑡 = 𝐴 cos 𝜔1 𝑡 + 𝜃 , 𝜆𝑇𝑜 ≤ 𝑡 ≤ 𝜆 + 1 𝑇𝑜 , for “1” 𝑠2 𝑡 = 𝐴 cos 𝜔2 𝑡 + 𝜃 , 𝜆𝑇𝑜 ≤ 𝑡 ≤ 𝜆 + 1 𝑇𝑜 , for “0” For coherent demodulation of the coherent FSK signal, the two frequencies are chosen so that the two signals are orthogonal: (𝜆+1)𝑇𝑜 𝑠1 (𝑡)𝑠2 (𝑡) 𝑑𝑡 = 0 𝜆𝑇𝑜 𝐴 cos 𝜔1 𝑡 + 𝜃 1 Frequency Synthesizer 𝜑𝐵𝐹𝑆𝐾 (𝑡) 2 𝐴 cos 𝜔2 𝑡 + 𝜃 44 Cont. ⇒ ⇒ (𝜆+1)𝑇𝑜 𝐴𝑐𝑜𝑠 𝜆𝑇𝑜 𝐴2 (𝜆+1)𝑇𝑜 2 𝜆𝑇𝑜 2𝜋𝑓1 𝑡 + 𝜃 𝐴𝑐𝑜𝑠 2𝜋𝑓2 𝑡 + 𝜃 𝑑𝑡 = 0 𝑐𝑜𝑠 2𝜋 𝑓1 + 𝑓2 𝑡 + 2𝜃 + 𝑐𝑜𝑠 2𝜋 𝑓1 − 𝑓2 𝑡 𝑑𝑡 = 0 𝐴2 ⇒ 2𝜋 𝑓 +𝑓 cos 2𝜃 𝑠𝑖𝑛 2𝜋 𝑓1 + 𝑓2 1 2 𝐴2 (1+𝜆)𝑇𝑜 𝑠𝑖𝑛 2𝜋 𝑓 − 𝑓 𝑡 | 1 2 𝜆𝑇𝑜 2𝜋 𝑓1 −𝑓2 (1+𝜆)𝑇𝑜 𝑡 + sin 2𝜃 𝑐𝑜𝑠 2𝜋 𝑓1 + 𝑓2 𝑡 |𝜆𝑇𝑜 + =0 This requires 2𝜋 𝑓1 + 𝑓2 𝑇𝑜 = 𝑛𝜋 and 2𝜋 𝑓1 − 𝑓2 𝑇𝑜 = 𝑚𝜋, 𝑛&𝑚 are integers. 𝑛+𝑚 4𝑇𝑜 𝑛−𝑚 𝑓2 = 4𝑇 𝑜 ⇒ 𝑓1 = ⇒ 2Δ𝑓 = 𝑓1 − 𝑓2 = 𝑚 2𝑇𝑜 45 Cont. Thus we conclude that for orthogonality 𝑓1 and 𝑓2 must be integer multiple of 1 4𝑇𝑜 and their difference must be integer multiples of 1 . 2𝑇𝑜 𝑓1 = 𝑓𝑐 + Δ𝑓 𝑓2 = 𝑓𝑐 − Δ𝑓 ⇒ 𝑓𝑐 = 𝑓1 +𝑓2 2 = 𝑛 2𝑇𝑜 Where 𝑓𝑐 is the nominal (or apparent) carrier frequency which must be an integer multiple of 1 2𝑇𝑜 for orthogonality. When the separation between 𝑓1 and 𝑓2 is chosen as 1 𝑇𝑜 , then the phase continuity will be maintained at bit transitions and the FSK is called Sunde ’s FSK. 46 Cont. As a matter of fact, if the separation is 𝛾 𝑇𝑜 , where 𝛾 is an integer, the phase of the coherent FSK signal is always continuous. Proof: at 𝑡 = 𝜆𝑇𝑜 , the phase of 𝑠1 (𝑡) is 2𝜋𝑓1 𝜆𝑇𝑜 + 𝜃 = 2𝜋(𝑓2 + 𝛾/𝑇𝑜 )𝜆𝑇𝑜 + 𝜃 = 2𝜋𝑓2 𝜆𝑇𝑜 + 2𝜋𝜆𝛾 + 𝜃 = 2𝜋𝑓2 𝜆𝑇𝑜 + 𝜃 (𝑀𝑜𝑑𝑢𝑙𝑜 2𝜋) which is exactly the phase of 𝑠2 (𝑡). Thus at 𝑡 = 𝜆𝑇𝑜 , if the input bit switches from 1 to 0, the new signal 𝑠2 (𝑡) will start at exactly the same amplitude where 𝑠1 (𝑡) has ended. The minimum separation for orthogonality between 𝑓1 and 𝑓2 is 1 2𝑇𝑜 . As we have just seen above, this separation cannot guarantee continuous phase. A particular form of FSK called minimum shift keying (MSK) not only has the minimum separation but also has continuous phase. 47 Cont. Figure (a) is an example of Sunde’s FSK waveform where bit 1 corresponds to a higher frequency 𝑓1 and bit 0 a lower 𝑓2 . Since 𝑓1 and 𝑓2 are multiples of 1 𝑇𝑜 , the ending phase of the carrier is the same as the starting phase, therefore the waveform has continuous phase at the bit boundaries. A coherent FSK waveform might have discontinuous phase at bit boundaries. Figure (b) is an example of such a waveform, where 𝑓1 = 9/4𝑇𝑜 , 𝑓2 = 6/4𝑇𝑜 , and the separation is 2Δ𝑓 = 3/4𝑇𝑜 . 48 PSD of Sunde’s FSK signal We expand the Sunde’s FSK signal as: 𝑔 𝑡 = 𝐴 cos 2𝜋 𝑓𝑐 + 𝑎𝜆 1 2𝑇𝑜 𝑡, 𝜆𝑇𝑜 ≤ 𝑡 ≤ 𝜆 + 1 𝑇𝑜 Where 𝑎𝜆 = ±1 ⇒ 𝑔 𝑡 = 𝐴 cos 𝑎𝜆 ⇒ 𝑔 𝑡 = 𝐴 cos ⇒ 𝑔𝐼 𝑡 = 𝐴 cos 𝜋𝑡 𝑇𝑜 𝜋𝑡 𝑇𝑜 ⇒ 𝑔𝑄 𝑡 = 𝐴𝑎𝜆 sin 𝜋𝑡 𝑇𝑜 cos 2𝜋𝑓𝑐 𝑡 − 𝐴 sin 𝑎𝜆 cos 2𝜋𝑓𝑐 𝑡 − 𝐴𝑎𝜆 sin 𝜋𝑡 𝑇𝑜 𝜋𝑡 𝑇𝑜 sin 2𝜋𝑓𝑐 𝑡 sin 2𝜋𝑓𝑐 𝑡 is independent of the data 𝜋𝑡 𝑇𝑜 is directly related to the data So the inphase and quadrature components are independent. 49 Cont. ⇒ 𝑔 𝑡 = 𝑔𝐼 (𝑡) cos 2𝜋𝑓𝑐 𝑡 − 𝑔𝑄 (𝑡) sin(2𝜋𝑓𝑐 𝑡) 𝑔 𝑡 = 𝑅𝑒 𝑔 𝑡 𝑒 𝑗2𝜋𝑓𝑐𝑡 𝑔 𝑡 = 𝑔𝐼 𝑡 + 𝑗 𝑔𝑄 (𝑡) 1 1 ⇒ 𝑆𝑔 𝑓 = 𝑆𝑔 𝑓 − 𝑓𝑐 + 𝑆𝑔 −𝑓 − 𝑓𝑐 4 4 Since the inphase component and the quadrature component of the FSK signal are independent of each other, the PSD for the complex envelope is the sum of the PSDs of these two components. ⇒ 𝑆𝑔 𝑓 = 𝑆𝐼 𝑓 + 𝑆𝑄 (𝑓) 𝑆𝐼 𝑓 = ℱ 𝐴 cos 𝜋𝑡 𝑇𝑜 2 = 𝐴2 4 𝛿 𝑓− 1 2𝑇𝑜 +𝛿 𝑓+ 1 2𝑇𝑜 50 Cont. 𝑔𝑄 𝑡 = 𝐴𝑎𝜆 sin 𝜋𝑡 𝑇𝑜 , 𝑎𝜆 = −1, +1 which are equally likely. We can look at it as a polar signal with pulse shape 𝑝 𝑡 = 𝐴 sin 𝑆𝑄 𝑓 = 𝑃(𝜔) 2 𝑇𝑜 ⇒ 𝑆𝑔 𝑓 = 𝐴2 4 = 1 𝑇𝑜 ℱ 𝐴 sin 𝛿 𝑓− 1 2𝑇𝑜 𝜋𝑡 𝑇𝑜 2 +𝛿 𝑓+ = 1 2𝑇𝑜 1 2𝐴𝑇𝑜 cos 𝜋𝑇𝑜 𝑓 𝑇𝑜 𝜋 1− 2𝑇𝑜 𝑓 2 + 𝜋𝑡 𝑇𝑜 2 2𝐴 cos 𝜋𝑇𝑜 𝑓 𝑇𝑜 𝜋 1− 2𝑇𝑜 𝑓 2 2 51 Orthogonality of Noncoherent BFSK 𝑠1 𝑡 = 𝐴 cos 𝜔1 𝑡 + 𝜃1 , 𝜆𝑇𝑜 ≤ 𝑡 ≤ 𝜆 + 1 𝑇𝑜 , for “1” 𝑠2 𝑡 = 𝐴 cos 𝜔2 𝑡 + 𝜃2 , 𝜆𝑇𝑜 ≤ 𝑡 ≤ 𝜆 + 1 𝑇𝑜 , for “0” 𝜃1 , 𝜃2 are initial phases at 𝑡 = 0. To simplify assume 𝜃1 = 0 and 𝜃2 = 𝜃 ⇒ 𝑠1 𝑡 = 𝐴 cos 𝜔1 𝑡 , 𝜆𝑇𝑜 ≤ 𝑡 ≤ 𝜆 + 1 𝑇𝑜 , for “1” ⇒ 𝑠2 𝑡 = 𝐴 cos 𝜔2 𝑡 + 𝜃 , 𝜆𝑇𝑜 ≤ 𝑡 ≤ 𝜆 + 1 𝑇𝑜 , for “0” (𝜆+1)𝑇𝑜 𝐴𝑐𝑜𝑠 2𝜋𝑓1 𝑡 𝐴𝑐𝑜𝑠 2𝜋𝑓2 𝑡 + 𝜃 𝑑𝑡 = 𝜆𝑇𝑜 (𝜆+1)𝑇𝑜 𝐴2 cos 𝜃 𝜆𝑇 𝑐𝑜𝑠 2𝜋𝑓1 𝑡 𝑐𝑜𝑠 2𝜋𝑓2 𝑡 𝑑𝑡 − 𝑜 (𝜆+1)𝑇𝑜 𝐴2 sin 𝜃 𝜆𝑇 𝑐𝑜𝑠 2𝜋𝑓1 𝑡 𝑠𝑖𝑛 2𝜋𝑓2 𝑡 𝑑𝑡 = 0 𝑜 𝑠𝑖𝑛 2𝜋 𝑓1 +𝑓2 𝑡 𝐴2 cos 𝜃 2𝜋 𝑓1 +𝑓2 𝑡 + 𝐴2 sin 𝜃 =0 ⇒ 0 𝑠𝑖𝑛 2𝜋 𝑓1 −𝑓2 𝑡 (𝜆+1)𝑇𝑜 + 2𝜋 𝑓1 −𝑓2 𝑡 𝜆𝑇𝑜 𝑐𝑜𝑠 2𝜋 𝑓1 +𝑓2 𝑡 𝑐𝑜𝑠 2𝜋 𝑓1 −𝑓2 𝑡 (𝜆+1)𝑇𝑜 + 2𝜋 𝑓1 +𝑓2 𝑡 2𝜋 𝑓1 −𝑓2 𝑡 𝜆𝑇𝑜 52 Cont. For arbitrary 𝜃, this requires that the sums inside the brackets be zero. This, in turn, requires that 2𝜋(𝑓1 + 𝑓2 )𝑇𝑜 = 𝑘𝜋 for the first term and 2𝜋(𝑓1 − 𝑓2 )𝑇𝑜 = 𝑙𝜋 for the second term in the first bracket, that 2𝜋(𝑓1 + 𝑓2 )𝑇𝑜 = 2𝑚𝜋 for the first term and 2𝜋(𝑓1 − 𝑓2 )𝑇𝑜 = 2𝑛𝜋 for the second term in the second bracket, where 𝑘, 𝑙, 𝑚, 𝑎𝑛𝑑 𝑛 are integers and 𝑘 > 𝑙, 𝑚 > 𝑛. The 𝑘𝜋 and 𝑙𝜋 cases are included in the 2𝑚𝜋 and 2𝑛𝜋 case, respectively. ⇒ 2𝜋(𝑓1 − 𝑓2 )𝑇𝑜 = 2𝑛𝜋 ⇒ 2𝜋(𝑓1 + 𝑓2 )𝑇𝑜 = 2𝑚𝜋 𝑛+𝑚 2𝑇𝑜 𝑚−𝑛 𝑓2 = 2𝑇𝑜 ⇒ 𝑓1 = ⇒ 2Δ𝑓 = 𝑓1 − 𝑓2 = 𝑛 𝑇𝑜 53 Cont. This is to say that for two noncoherent FSK signals to be orthogonal, the two frequencies must be integer multiples of 1/2𝑇𝑜 and their separation must be a multiple of 1/𝑇𝑜 . When 𝑛 = 1, the separation is the 1/𝑇𝑜 , which is the minimum. Comparing with a coherent FSK case, the separation of noncoherent FSK is double that of FSK. Thus more system bandwidth is required for noncoherent FSK for the same symbol rate. 54 Continuous phase FSK(CPFSK): 𝜑 𝑡 = 𝐴 𝑝 𝑡 cos 𝜔𝑐 𝑡 + 𝜃(𝑡) , 𝜆𝑇𝑜 ≤ 𝑡 ≤ 𝜆 + 1 𝑇𝑜 Where θ(t) is a continuous function of time 𝜃 𝑡 = 𝜃 0 + ∆𝜔𝑡 , "1" , "0" 𝜃 𝑡 = 𝜃 0 − ∆𝜔𝑡 Where θ(0) = phase at (t = 0) ⇒ 𝜃 𝑡 = 𝜃 0 ± 2𝜋∆𝑓𝑡 𝜃 𝑡 = 𝜃 0 ± 2𝜋( 𝑛𝑅𝑜 2 )𝑡 𝜃 𝑡 = 𝜃 0 ± 𝜋𝑛𝑅𝑜 𝑡 𝜃 𝑡 = 𝜃 0 ± 𝜋𝑛 𝑡 𝑇𝑜 , 0 ≤ 𝑡 ≤ 𝑇𝑜 CPFSK (Cont.) → 𝜑 𝑡 = 𝐴 𝑝 𝑡 cos 𝜔1 𝑡 + 𝜃(0) → "1“ & 𝜑 𝑡 = 𝐴 𝑝 𝑡 cos 𝜔2 𝑡 + 𝜃(0) → "0" 𝑛 𝑓1 = 𝑓𝑐 + & 2𝑇𝑜 1 → 𝑓𝑐 = (𝑓1 + 𝑓2 ) 2 𝑓2 = 𝑓𝑐 − 𝑛 2𝑇𝑜 → 𝑛 = 𝑇0 (𝑓1 − 𝑓2 ) Where n is called the deviation ratio 𝜃 𝑡 −𝜃 0 =± 𝜋 𝑛𝑡 𝑇𝑜 At 𝑇𝑜 : 𝜃(𝑇𝑜 ) − 𝜃 0 = ±𝜋𝑛 → sending "1" increase the phase by πn radian → sending "0" decrease the phase by πn radian Example: Data : 110100 𝜃 𝑡 − 𝜃(0) 2𝜋𝑛 𝜋𝑛 𝑡 2𝑇𝑜 −2𝑇𝑜 4𝑇𝑜 6𝑇𝑜 −𝜋𝑛 −2𝜋𝑛 Phase Tree CPFSK (Cont.) Let 𝑛 = 1 2 → 𝜃(𝑡) − 𝜃 0 = 𝜋 ± 𝑡 2𝑇𝑜 𝜋 =± 2 𝑎𝑡 𝑡 = 𝑇𝑜 → 𝜃 𝑡 −𝜃 0 𝑎𝑡 𝑡 = 2𝑇𝑜 → 𝜃 𝑡 − 𝜃 0 = ±𝜋 𝑎𝑡 𝑡 = 3𝑇𝑜 → 𝜃 𝑡 −𝜃 0 = 3𝜋 ± 2 = 𝜋 ∓ 2 𝜃 𝑡 − 𝜃(0) Example: Data : 1101100 𝜋 𝜋/2 2𝑇𝑜 −2𝑇𝑜 4𝑇𝑜 𝑡 6𝑇𝑜 −𝜋/2 −𝜋 Phase Trellis Minimum Shift Keying (MSK) what is the min Δf? → minimum shift keying (MSK); it is a continuous phase FSK → CPFSK 𝑛= 1 2 → ∆𝑓 = 𝑅𝑜 4 → ∆𝑓 = 1 4𝑇𝑜 How? Polar min BW is 𝑅𝑜 2 → (zero ISI or Duobinary pulse) Duobinary − 𝑅𝑜 2 𝑅𝑜 2 f 𝐵𝑊 = 1.5𝑅𝑜 𝑅𝑜 2 𝑅𝑜 𝑓2 − 2 𝑓2 = 𝑓𝑐 − ∆𝑓 𝑓𝑐 𝑓1 = 𝑓𝑐 + ∆𝑓 𝑅𝑜 𝑓1 + 2 f BFSK Demodulation: 1. Coherent: 62 2. Non-coherent: 𝜔1 𝜔2 𝜔 𝜔 63 Constellation of BFSK 𝜓2 (𝑡) 𝑠2 𝑠1 𝜓1 (𝑡) 64 Binary PSK (BPSK): 𝜑𝐵𝑃𝑆𝐾 𝑡 = 𝑦 𝑡 . cos(𝜔𝑐 𝑡) → DSB-SC y(t) : polar NRZ 65 Polar NRZ 1 0 0 1 1 0 A -A 180° 180° 180° 66 𝑃(𝜔) 𝑆𝑌 𝜔 = 𝑇 2 𝑃 𝜔 = 𝑇0 2 ∞ ℛ𝑛 𝑒 −𝑗𝑛𝜔𝑇 𝑛=−∞ For polar line code 𝑃 𝜔 =ℱ 𝑝 𝑡 ] 𝑝 𝑡 =𝐴∙ 𝑡−𝑇0 2 𝑟𝑒𝑐𝑡( ) 𝑇0 (NRZ rect) 𝑝(𝑡) 𝑃 𝜔 𝜔𝑇0 −𝑗𝜔𝑇0 2 = 𝐴 ∙ 𝑇0 𝑠𝑖𝑛𝑐( )𝑒 2 𝜔𝑇0 2 2 2 2 𝑃 𝜔 = 𝐴 ∙ 𝑇0 𝑠𝑖𝑛𝑐 ( ) 2 𝑆𝑌 𝜔 = 𝐴2 ∙ 𝑇0 𝑠𝑖𝑛𝑐 2 ( 𝑇0 2 t 𝑇0 𝜔𝑇0 ) 2 67 𝑆𝐵𝑃𝑆𝐾 𝜔 𝐴2 𝑇𝑜 4 𝜔𝑐 − 2𝜋 𝑓𝑐 − 2𝑅0 𝑓𝑐 − 𝑅0 𝑇0 𝜔𝑐 𝜔 𝜔𝑐 + 2𝜋 𝑇 0 𝑓 𝐸𝐵 = 2𝑅𝑜 𝑓𝑐 + 𝑅0 𝑓𝑐 + 2𝑅0 68 BPSK (Cont.) +𝐴 cos 𝜔𝑐 𝑡 ⇒ "1" , 𝜆𝑇𝑜 ≤ 𝑡 ≤ (𝜆 + 1)𝑇𝑜 −𝐴 cos 𝜔𝑐 𝑡 ⇒ Acos(𝜔𝑐 𝑡 + 𝜋) ⇒ "0" , 𝜆𝑇𝑜 ≤ 𝑡 ≤ 𝜆 + 1 𝑇𝑜 ⇒ 𝜑𝑃𝑆𝐾 𝑡 = 𝐴 𝑐𝑜𝑠 𝜔𝑐 𝑡 + 𝜃𝜆 , 𝜆𝑇𝑜 ≤ 𝑡 ≤ 𝜆 + 1 𝑇𝑜 ⇒ (PM) 𝜃𝜆 constant with respect to t ⇒ instanteneous frequency is constant 𝜑𝑃𝑆𝐾 𝑡 = 𝐴 𝑐𝑜𝑠𝜃𝜆 cos 𝜔𝑐 𝑡 − 𝐴 𝑠𝑖𝑛𝜃𝜆 sin(𝜔𝑐 𝑡) 𝜑𝑃𝑆𝐾 𝑡 = 𝑎𝜆 cos 𝜔𝑐 𝑡 + 𝑏𝜆 sin 𝜔𝑐 𝑡 ⇒ 𝑄𝐴𝑀 𝑎𝜆 = 𝐴 𝑐𝑜𝑠𝜃𝜆 𝑏𝜆 = −𝐴 𝑠𝑖𝑛𝜃𝜆 In binary 𝜃𝜆 = 0, π} ⇒ 𝑎𝜆 = ±𝐴 , 𝑏𝜆 = 0 ⇒ 𝜑𝐵𝑃𝑆𝐾 𝑡 = ±𝐴 cos(𝜔𝑐 𝑡) 69 Constellation of BPSK 𝜓2 (𝑡) 𝑠2 𝑠1 𝜓1 (𝑡) We observe that 𝜓2 (𝑡) has nothing to do with signals. Hence, only one basis function is sufficient to represent the signals 70 BPSK Demodulation: Coherent: 𝑦(𝑡) 𝜑 𝑡 = y(t)cos(𝜔𝑐 𝑡) LPF cos(𝜔𝑐 𝑡) Cannot use ED 71 Differential BPSK (DBPSK): - To use non-coherent detection. Receiver detect the relative phase change between successive modulation phases θk & θk-1 𝜃𝑘 − 𝜃𝑘−1 = 0 ⇒ "0“ 𝜃𝑘 − 𝜃𝑘−1 = 𝜋 ⇒ "1“ 72 DBPSK(Cont.) 𝐼𝑘 = "0", "1" 𝑎𝑘 = 2𝑑𝑘 − 1 = +𝐴, −𝐴 𝑑𝑘 = 𝐼𝑘 ⊕ 𝑑𝑘−1 = 0, 1 Delay To Polar generator 𝜑(𝑡) ± Acos(𝜔𝑐 𝑡) cos(𝜔𝑐 𝑡) ⊕ 𝑋𝑂𝑅 𝑜𝑟 𝑀𝑜𝑑𝑢𝑙𝑜 2 𝑎𝑑𝑑𝑒𝑟 73 𝐼𝑘 1 0 0 1 1 0 𝑑𝑘 0 1 1 1 0 1 1 +𝐴 +𝐴 +𝐴 −𝐴 +𝐴 +𝐴 NRZI: Non return to zero inverted 𝑎𝑘 +𝐴 −𝐴 74 DBPSK (Cont.) Sending "0" will keep the previous phase. Sending "1" change the previous phase by π. ⇒ when "0" is Tx : both the current & previous pulses will be: +A cos(ωc t) or −A cos(ωc t) ⇒ when "1" is Tx : the current pulse & the previous pulse will differ by the sign 𝑐𝑢𝑟𝑟𝑒𝑛𝑡 = ± cos 𝜔𝑐 𝑡 , 𝑝𝑟𝑒𝑣𝑖𝑜𝑢𝑠 = ∓cos(𝜔𝑐 𝑡) 75 DBPSK Demodulator 𝑙(𝑡) 𝑟(𝑡) 𝜑(𝑡) LPF Decision 𝑙 𝑡 > 0 ⇒ "0" 𝑙 𝑡 < 0 ⇒ "1" Delay To 𝐴2 (1 2 ⇒"0": 𝑟 𝑡 = 𝜑 𝑡 . 𝜑 𝑡 − 𝑇𝑜 = + cos(2𝜔𝑐 𝑡)) 𝐴2 𝑙 𝑡 = 2 −𝐴2 ⇒“1": 𝑟 𝑡 = 𝜑 𝑡 . 𝜑 𝑡 − 𝑇𝑜 = (1 + cos(2𝜔𝑐 𝑡)) 2 −𝐴2 𝑙 𝑡 = 2 76 M-ary Digital Modulation To increase data rate or Reduce the Digital signal Bandwidth. 𝑀 = 2𝑘 , k: # of bits , M: # of symbols (pulses) 𝑘 = 𝑙𝑜𝑔2 𝑀 k bits are assigned 1 of the M symbols. Send one symbol in 𝑇𝑀 . 𝑇𝑀 is the symbol duration: # of symbols per sec is 𝑅𝑀 = 1 𝑇𝑀 Baud rate [Baud] 77 M-ary ASK: Like Binary ASK it is a special case of AM signal with Digital message signal. For AM with 𝒎 = 𝟏: 𝜑𝑙 𝑡 = 𝑙 ∙ 𝑝 𝑡 cos(𝜔𝑐 𝑡) , 𝑙 = 0,1,2, … . , (𝑀 − 1) , λ𝑇𝑀 ≤ 𝑡 ≤ (𝜆 + 1)𝑇𝑀 Where 𝑝 𝑡 is a pulse shape. e.g. M=4, k=2, 𝑡−𝑇𝑀 𝑝 𝑡 𝑖𝑠 𝑟𝑒𝑐𝑡 𝑇 𝑀 Bits Amp. 00 0 01 +1 10 +3 11 +2 2 Demodulation: Non-Coherent ED Coherent Constellation 𝑠0 𝑠1 ….. 𝑠𝑀−1 𝜓1 (𝑡) 78 M-ary ASK (Cont.) For AM with 𝒎 < 𝟏: 𝜑𝑙 𝑡 = 𝑙 ∙ 𝑝 𝑡 cos(𝜔𝑐 𝑡) , 𝑙 = 1,2, … . , 𝑀 , λ𝑇𝑀 ≤ 𝑡 ≤ (𝜆 + 1)𝑇𝑀 Where 𝑝 𝑡 is a pulse shape. e.g. M=4, k=2, 𝑡−𝑇𝑀 𝑝 𝑡 𝑖𝑠 𝑟𝑒𝑐𝑡 𝑇 𝑀 Bits Amp. 00 +1 01 +2 10 +4 11 3 2 Demodulation: Non-Coherent ED Coherent Constellation 𝑠1 𝑠2 ….. 𝑠𝑀 𝜓1 (𝑡) 79 M-ary ASK (Cont.) For AM with 𝒎 > 𝟏: 𝜑𝑙 𝑡 = ±𝑙 ∙ 𝑝 𝑡 cos(𝜔𝑐 𝑡) , 𝑙 = 1,3,5, … . , (𝑀 − 1) , λ ≤ 𝑡 ≤ (𝜆 + 1)𝑇𝑀 Where 𝑝 𝑡 is a pulse shape. e.g. M=4, k=2, 𝑡−𝑇𝑀 𝑝 𝑡 𝑖𝑠 𝑟𝑒𝑐𝑡 𝑇 2 Demodulation: Coherent only 𝑀 Bits Amp. 00 -3 01 -1 10 +3 11 +1 Constellation ….. ….. 𝜓1 (𝑡) 80 M-ary ASK (Cont.) PSD for 𝑚 > 1 and M=4: 𝑃(𝜔) 𝑆𝑌 𝜔 = 𝑇𝑀 5𝑃 𝜔 = 𝑇𝑀 ∞ 2 ℛ𝑛 𝑒 −𝑗𝑛𝜔𝑇𝑀 𝑛=−∞ 2 𝑃 𝜔 =ℱ 𝑝 𝑡 ] 𝑝 𝑡 = 𝑡−𝑇𝑀 2 𝑟𝑒𝑐𝑡( ) 𝑇𝑀 (NRZ rect) 𝜔𝑇𝑀 ) 2 𝑆𝑌 𝜔 = 5 ∙ 𝑇𝑀 𝑠𝑖𝑛𝑐 2 ( 81 𝑆𝑀𝐴𝑆𝐾 𝜔 5𝑇𝑀 4 𝜔𝑐 − 2𝜋 −2𝑅M −𝑅M 𝑇𝑀 𝜔𝑐 𝜔 𝜔𝑐 + 2𝜋 𝑇 𝑀 𝑓 𝐸𝐵 = 2𝑅𝑀 𝑅M 2𝑅M 𝜌 =? 82 M-ary FSK (Multi tone Signaling): Each pulse is transmitted as: 𝜑𝑖 𝑡 = 𝐴 cos(𝜔𝑖 𝑡) , 𝜆𝑇𝑀 ≤ 𝑡 ≤ 𝜆 + 1 𝑇𝑀 , 𝑖 = 1,2, … , 𝑀 𝜑𝑖 𝑡 = 𝐴 cos(2𝜋𝑓𝑖 𝑡) , 𝜆𝑇𝑀 ≤ 𝑡 ≤ (𝜆 + 1)𝑇𝑀 𝑖 = 1,2, … , 𝑀 FSK is a special case of FM. 83 M-ary FSK (Cont.) One way to select 𝑓𝑖 is: 𝑓𝑖 = 𝑓1 + 𝑖 − 1 𝛿𝑓 , 𝑖 = 1,2,3, … , 𝑀 Lower frequency f1 Upper frequency f2 f3 fc 𝑓𝑀 = 𝑓1 + (𝑀 − 1)𝛿𝑓 2∆𝑓 ∆𝑓: Frequency deviation 84 M-ary FSK (Cont.) 2∆𝑓 = 𝑓𝑀 − 𝑓1 = (𝑀 − 1)𝛿𝑓 ∆𝑓 = (𝑀−1)𝛿𝑓 2 As ∆𝑓 increase Bandwidth increase As ∆𝑓 decrease Bandwidth decrease For fixed value of M, ∆𝑓 is controlled through 𝛿𝑓. 𝑡𝑜𝑜 ℎ𝑖𝑔ℎ, 𝐵𝑊 𝑖𝑠 ℎ𝑖𝑔ℎ 𝛿𝑓 ⇒ 𝑡𝑜𝑜 𝑙𝑜𝑤, 𝑐𝑎𝑛 𝑛𝑜𝑡 𝑑𝑖𝑠𝑡𝑖𝑛𝑔𝑢𝑖𝑠ℎ 𝑝𝑢𝑙𝑠𝑒𝑠 85 M-ary FSK (Cont.) Then the question is: What is the minimum value of 𝛿𝑓 such that we can distinguish a pulse from another? The answer is through (Orthogonality principle). The signal set 𝜑𝑖 (𝑡) is orthogonal set over the time interval 𝑎, 𝑏 if the inner product of any two signals in this set is zero. 86 M-ary FSK (Cont.) 𝐸𝑛 , 𝑖𝑓 𝑚 = 𝑛 𝜑𝑚 (𝑡), 𝜑𝑛 (𝑡) = 𝑡 𝜑𝑛 𝑡 𝑑𝑡 = 0, 𝑖𝑓 𝑚 ≠ 𝑛 𝐸𝑛 is the energy of 𝜑𝑛 (𝑡) over the interval 𝑎, 𝑏 If 𝐸𝑖 = 1 for every signal in the set then the set is called an orthonormal signal set. To transform an orthogonal set into an orthonormal set we divide each signal by the square root of its energy ( 𝐸𝑖 ). The set 𝑏 𝜑 𝑎 𝑚 𝜑𝑖 (𝑡) 𝐸𝑖 ∗ is orthonormal signal set. 87 M-ary FSK (Cont.) ∗ 𝑏 𝜑𝑚 𝑡 𝜑𝑛 𝑡 𝑎 𝐸𝑚 𝐸𝑛 𝑑𝑡 = 𝐸𝑛 𝐸𝑛 = 1, 𝑖𝑓 𝑚 = 𝑛 0, 𝑖𝑓 𝑚 ≠ 𝑛 Let 𝜑𝑖 𝑡 = 𝐴 cos(2𝜋𝑓𝑖 𝑡) for 𝑖 = 1,2, … , 𝑀 be a signal set defined of the interval 0, 𝑇𝑀 . Then 𝜑𝑚 𝑡 = 𝐴 cos(2𝜋𝑓𝑚 𝑡) and 𝜑𝑛 𝑡 = 𝐴 cos(2𝜋𝑓𝑛 𝑡) To be orthogonal: 𝑇𝑀 2 𝐴 0 cos(2𝜋𝑓𝑚 𝑡) cos(2𝜋𝑓𝑛 𝑡) 𝑑𝑡 = 0, 𝑓𝑜𝑟 𝑚 ≠ 𝑛. 88 M-ary FSK (Cont.) 𝐴2 = 2 𝐴2 = 2 𝑇𝑀 cos(2𝜋 𝑓𝑚 + 𝑓𝑛 𝑡) + cos(2𝜋 𝑓𝑚 − 𝑓𝑛 𝑡) 𝑑𝑡 0 𝑇𝑀 𝑇𝑀 cos( 2𝜋 𝑓𝑚 + 𝑓𝑛 𝑡)𝑑𝑡 + 0 cos(2𝜋 𝑓𝑚 − 𝑓𝑛 𝑡)𝑑𝑡 0 Zero 𝐴2 sin(2𝜋 𝑓𝑚 + 𝑓𝑛 𝑇𝑀 sin(2𝜋 𝑓𝑚 − 𝑓𝑛 𝑇𝑀 = 𝑇 ∙ + 𝑇𝑀 2 𝑀 2𝜋 𝑓𝑚 + 𝑓𝑛 𝑇𝑀 2𝜋(𝑓𝑚 − 𝑓𝑛 )𝑇𝑀 Since practically 𝑓𝑚 + 𝑓𝑛 𝑇𝑀 is very high. 89 M-ary FSK (Cont.) sin 2𝜋 𝑓𝑚 −𝑓𝑛 𝑇𝑀 2𝜋 𝑓𝑚 −𝑓𝑛 =0 , 𝑚≠𝑛 sin( 2𝜋 𝑓𝑚 − 𝑓𝑛 𝑇𝑀 ) = 0 𝑓𝑚 = 𝑓1 + (𝑚 − 1)𝛿𝑓 ⇒ 𝑓𝑛 = 𝑓1 + (𝑛 − 1)𝛿𝑓 𝑓𝑚 − 𝑓𝑛 = (𝑚 − 𝑛)𝛿𝑓 ⇒ sin(2𝜋 𝑚 − 𝑛 𝛿𝑓. 𝑇𝑀 = 0 ⇒ 2𝜋 𝑚 − 𝑛 𝛿𝑓. 𝑇𝑀 = 𝜋 𝑜𝑟 𝑚𝑢𝑙𝑡𝑖𝑝𝑙𝑒 𝑜𝑓 𝑖𝑡 ⇒ 𝛿𝑓 = ⇒ 𝛿𝑓 = 1 minimum when 𝑚 2𝑇𝑀 (𝑚−𝑛) 1 𝑅 = 𝑀 (min) 2𝑇𝑀 2 −𝑛 =1 e.g. Assume 3 orthogonal pulses 𝜓2 (𝑡) 𝑠2 𝑠1 𝜓1 (𝑡) (𝑀 − 1) ⇒ ∆𝑓 = 4𝑇𝑀 & 𝑓𝑖 = 𝑓1 + (𝑖−1) 2𝑇𝑀 𝑠3 𝜓3 (𝑡) , 𝑖 = 1,2,3, … . , 𝑀 90 M-ary FSK (Cont.) 𝜑𝑖 𝑡 = 𝐴 cos 2𝜋 𝑓1 + 𝑓4 𝑓3 𝑓2 𝑓3 (𝑖−1) 2𝑇𝑀 𝑓4 𝑡 , 𝑖 = 1,2, … … . . 𝑀 𝑓1 A -A Bits Freq. 00 𝑓1 01 𝑓2 10 𝑓3 11 𝑓4 91 Each pulse can be visualized as an OOK signal. Σ 𝑆𝑀𝐹𝑆𝐾 𝜔 If 𝑧 𝑡 = 𝑥1 𝑡 + 𝑥2 𝑡 + ⋯ ⋯ ⋯ + 𝑥𝑀 𝑡 And 𝑥𝑖 𝑡 , 𝑥𝑗 𝑡 are mutually orthogonal for i,j=1,2,…,M ⇒ 𝑆 𝜔 = 𝑆𝑥1 𝜔 + 𝑆𝑥2 𝜔 + ⋯ ⋯ ⋯ 𝑆𝑥𝑀 (𝜔) 2𝑅𝑀 2𝑅𝑀 ………………………………. 𝜔1 𝑓1 − 𝑅𝑜 𝑓 1 𝜔1 + 2π𝑅𝑀 𝜔𝑀 𝜔𝑀 + 2π𝑅𝑀 𝜔2 𝜔 𝑓 𝑓𝑀 + 𝑅𝑀 𝑓2 2∆𝑓 + 2𝑅𝑀 Orthogonal M-ary FSK 𝑓𝑖 = 𝑓1 + 𝑖 − 1 𝛿𝑓 , 𝑖 = 1,2,3, … , 𝑀 Lower frequency f1 Upper frequency f2 𝑅𝑀 2 f3 fc 𝑓𝑀 = 𝑓1 + (𝑀 − 1)𝛿𝑓 2∆𝑓 94 M-ary FSK (Cont.) Using Carson’s Rule: 𝐵𝑊 = 2∆𝑓 + 2𝐵 Assume the base band signal y(t) is polar NRZ → Bandwidth of y t is 𝐵 = 1 𝑇𝑀 = 𝑅𝑀 𝐵𝑊 𝑜𝑓 𝑀𝐹𝑆𝐾 = 2∆𝑓 + 2𝑅𝑀 = 𝑀−1 2𝑇𝑀 + 2 𝑇𝑀 = 𝑀+3 2𝑇𝑀 Increasing M will increase the BW linearly but the power will remains constant. For orthogonal BW = 2∆𝑓 + 𝑅𝑀 = 𝑀+1 2𝑇𝑀 ≈ 𝑀 2𝑇𝑀 95 M-ary FSK Demodulation: -Coherent cos(𝜔1 𝑡) 𝑡 = n𝑇𝑀 LPF sampler cos(𝜔2 𝑡) 𝑡 = n𝑇𝑀 LPF sampler 𝜑𝑀𝐹𝑆𝐾 𝑡 …….. …….. …….. cos(𝜔𝑀 𝑡) Comparator Select Largest 𝑡 = n𝑇𝑀 LPF sampler 96 M-ary FSK Demodulation: -Non-Coherent 𝐻𝑖 (𝜔) is a bandpass filter centered at 𝜔𝑖 𝑡 = n𝑇𝑀 𝐻1 (𝜔) ED sampler 𝑡 = n𝑇𝑀 𝐻2 (𝜔) ED sampler 𝜑𝑀𝐹𝑆𝐾 𝑡 Comparator Select Largest …….. …….. …….. 𝑡 = n𝑇𝑀 𝐻𝑀 (𝜔) ED sampler 97 M-ary PSK: Each pulse is transmitted as: 𝜑𝑖 𝑡 = 𝐴 cos(𝜔𝑐 𝑡 + 𝜃𝑖 ) , 𝜆𝑇𝑀 ≤ 𝑡 ≤ 𝜆 + 1 𝑇𝑀 , 𝑖 = 1,2, … , 𝑀 𝜑𝑖 𝑡 = 𝐴 cos(2𝜋𝑓𝑐 𝑡 + 𝜃𝑖 ) , 𝜆𝑇𝑀 ≤ 𝑡 ≤ (𝜆 + 1)𝑇𝑀 𝑖 = 1,2, … , 𝑀 Where 𝑓𝑐 = 𝑛 𝑇𝑀 , n:integer Tc 𝑇𝑀 = 𝑛𝑇𝑐 , 𝑇𝑐 is the period of the carrier. TM 98 M-ary PSK (Cont.) One way to choose 𝜃𝑖 is: 𝜃𝑖 = 𝜃𝑜 + 2𝜋 (𝑖 𝑀 − 1), 𝑖 = 1,2, … , 𝑀 𝜃𝑜 is an initial phase. 𝜑𝑖 𝑡 = 𝐴 cos(2𝜋𝑓𝑐 𝑡 + 𝜃𝑖 ) = 𝐴 cos 𝜃𝑖 cos 𝜔𝑐 𝑡 − 𝐴 sin 𝜃𝑖 sin 𝜔𝑐 𝑡 = 𝑎𝑖 cos 𝜔𝑐 𝑡 − 𝑏𝑖 sin(𝜔𝑐 𝑡) DSB-SCASK DSB-SCASK 𝑎𝑖 = 𝐴 cos(𝜃𝑖 ) : The In-phase component (I). 𝑏𝑖 = 𝐴 sin(𝜃𝑖 ) : The Quadrature component (Q). 99 M-ary PSK (Cont.) 𝜑𝑖 𝑡 = 𝑎𝑖 cos 𝜔𝑐 𝑡 − 𝑏𝑖 sin(𝜔𝑐 𝑡) The cos(𝜔𝑐 𝑡) and sin(𝜔𝑐 𝑡) are orthogonal. PSD of MPSK = PSD of 𝑎𝑖 cos 𝜔𝑐 𝑡 +PSD of 𝑏𝑖 sin(𝜔𝑐 𝑡) PSD of s(t) = 𝑚(𝑡) sin(𝜔𝑐 𝑡) is: 1 𝑆𝑠 𝜔 = 𝑆𝑚 𝜔 − 𝜔𝑐 + 𝑆𝑚 (𝜔 + 𝜔𝑐 ) 4 PSD of MPSK = 2 ∙ 𝑃𝑆𝐷 𝑜𝑓 𝐴𝑆𝐾 100 𝑆𝑀𝑃𝑆𝐾 𝜔 𝜔𝑐 − 2𝜋 𝑓𝑐 − 2𝑅𝑀 𝑓𝑐 − 𝑅𝑀 𝑇𝑀 𝜔𝑐 𝜔 𝜔𝑐 + 2𝜋 𝑇 𝑀 𝑓 𝐸𝐵 = 2𝑅𝑀 𝑓𝑐 + 𝑅𝑀 𝑓𝑐 + 2𝑅𝑀 101 M-ary PSK (Cont.) 𝑟= 𝜃𝑖 = 𝑎𝑖 2 + 𝑏𝑖 2 = 𝐴2 𝑐𝑜𝑠 2 𝜃𝑖 + 𝐴2 𝑠𝑖𝑛2 𝜃𝑖 = 𝐴 −1 𝑏𝑖 𝑡𝑎𝑛 𝑎𝑖 Let 𝑧 = 𝑎𝑖 + 𝑗𝑏𝑖 called the complex envelope (the equivalent low-pass signal of 𝜑𝑖 (𝑡)). 𝜓2 (𝑡) 𝜑𝑖 𝑡 = 𝑅𝑒 (𝑎𝑖 + 𝑗𝑏𝑖 )𝑒 𝑗𝜔𝑐 𝑡 = 𝑅𝑒 (𝑎𝑖 +𝑗𝑏𝑖 )(cos 𝜔𝑐 𝑡 + 𝑗 sin 𝜔𝑐 𝑡) 𝑏𝑖 𝑟 𝜃𝑖 𝑎𝑖 = 𝑅𝑒 𝑎𝑖 cos 𝜔𝑐 𝑡 + 𝑗𝑎𝑖 sin 𝜔𝑐 𝑡 + 𝑗𝑏𝑖 cos 𝜔𝑐 𝑡 − 𝑏𝑖 sin 𝜔𝑐 𝑡 = 𝑎𝑖 cos 𝜔𝑐 𝑡 − 𝑏𝑖 sin 𝜔𝑐 𝑡 𝜓1 (𝑡) 102 M-ary PSK (Cont.) ⇒ 𝑧 = 𝑎𝑖 + 𝑏𝑖 = 𝑟 ⇒ ∠𝑧 = tan 𝑏𝑖 𝑎𝑖 = 𝜃𝑖 103 Constellation of 8-ary PSK 𝜓2 (𝑡) All the points will stay on a circle 𝑀=8 𝑠3 𝑠2 𝑠4 𝑠1 𝑠5 𝜓1 (𝑡) 𝑠6 𝑠8 𝑠7 104 M-ary PSK (Cont.) Demodulation: Coherent only 𝑟 𝑡 = 𝐴 2 × 2 cos 𝜃𝑖 + cos(2𝜔𝑐 𝑡 + 𝜃𝑖 ) 𝑙 𝑡 = 𝐴𝑐𝑜𝑠(𝜃𝑖 ) 𝜑𝑖 𝑡 = Acos(𝜔𝑐 𝑡 + 𝜃𝑖 ) 𝑙(𝑡) 𝑟(𝑡) LPF 2cos(𝜔𝑐 𝑡) Why? 105 Quadriphase Shift keying (QPSK): Known also as Quaternary PSK. 𝑀 = 4 ⇒ 𝑘 = 2. Each pulse is transmitted as: 𝜑𝑖 𝑡 = 𝐴 cos(𝜔𝑐 𝑡 + 𝜃𝑖 ) , 𝜆𝑇𝑀 ≤ 𝑡 ≤ 𝜆 + 1 𝑇𝑀 , 𝑖 = 1,2,3,4 𝜋 4 2𝜋 𝑖−1 𝜃𝑖 = + , 𝑖 = 1,2,3,4 4 𝜋 3𝜋 5𝜋 7𝜋 So 𝜃1 = , 𝜃2 = , 𝜃3 = , 𝜃4 = 4 4 4 4 106 Constellation of QPSK 𝜓2 (𝑡) 𝑀=4 dibit ≜ two bits Bits 00 01 10 11 Phase 5𝜋 4 3𝜋 4 7𝜋 4 𝜋 4 dibit 𝑠2 01 𝑠1 11 𝜓1 (𝑡) 𝑠3 00 𝑠4 10 How to assign a code word for the neighbor pulse? Gray code 107 QPSK (Cont.) 𝜋 4 𝜋 4 5𝜋 4 3𝜋 4 7𝜋 4 108 QPSK (Cont.) Gray Code: Assume 𝑏1 𝑏2 𝑏3 … 𝑏𝑘 is a 𝑘 bit code word 𝑏𝑖 = "0", "1" , 𝑖 = 1,2, … , 𝑘 And 𝑔1 𝑔2 𝑔3 … 𝑔𝑘 is the corresponding Gray code work. 𝑔𝑖 = "0", "1" , 𝑖 = 1,2, … , 𝑘 Then: 𝑔1 = 𝑏1 𝑔𝑚 = 𝑏𝑚 ⊕ 𝑏𝑚−1 , 𝑚 ≥ 2 Where ⊕ is XOR 109 QPSK (cont.) ex: 𝑘=3 𝑘=2 b g b g 000 000 00 00 001 001 01 01 010 011 10 11 011 010 11 10 100 110 101 111 110 101 111 100 110 QPSK (Cont.) 𝑎1 = 𝐴 𝑐𝑜𝑠 𝜋 4 𝑎2 = 𝐴 𝑐𝑜𝑠 3𝜋 4 = 𝐴 2 , =− 𝐴 2 𝑎3 = 𝐴 𝑐𝑜𝑠 5𝜋 4 =− 𝐴 2 𝑎4 = 𝐴 𝑐𝑜𝑠 7𝜋 4 𝐴 2 = , , , 𝑏1 = 𝐴 sin 𝜋 4 𝑏2 = 𝐴 sin 3𝜋 4 𝑏3 = 𝐴 sin 5𝜋 4 𝑏4 = 𝐴 sin 7𝜋 4 = = 𝐴 2 𝐴 2 =− 𝐴 2 =− 𝐴 2 𝐴 𝑚𝑎𝑝𝑝𝑒𝑑 𝑡𝑜 "0" 2 𝐴 + 𝑚𝑎𝑝𝑝𝑒𝑑 𝑡𝑜 "1” 2 Let − And 111 QPSK (Cont.) 𝜓2 (𝑡) b g 00 00 01 01 10 11 11 10 𝒂𝒊 𝐴 − 2 𝐴 − 2 𝐴 + 2 𝐴 + 2 𝒃𝒊 𝐴 − 2 𝐴 + 2 𝐴 + 2 𝐴 − 2 𝑀=4 𝑠2 01 𝑠1 11 𝜓1 (𝑡) 𝑠3 00 𝑠4 10 112 QPSK (Cont.) 0 0 𝑎𝑖 cos(𝜔𝑐 𝑡) 1 1 1 1 0 1 1 0 BASK ≡BPSK 𝑏𝑖 sin(𝜔𝑐 𝑡) BASK ≡BPSK 𝜑𝑖 𝑡 = 𝑎𝑖 cos 𝜔𝑐 𝑡 − 𝑏𝑖 sin(𝜔𝑐 𝑡) 113 QPSK (Cont.) PSD: 𝑆𝑄𝑃𝑆𝐾 𝜔 = 𝑆𝐵𝑃𝑆𝐾 𝜔 + 𝑆𝐵𝑃𝑆𝐾 (𝜔) = 𝐴2 𝑇𝑀 2 𝜔−𝜔𝑐 𝑇𝑀 2 𝑠𝑖𝑛𝑐 2 (𝜔+𝜔𝑐 )𝑇𝑀 2 + 𝑠𝑖𝑛𝑐 ( ) 2 𝑆𝑄𝑃𝑆𝐾 𝜔 𝜔𝑐 − 2𝜋 𝑓𝑐 − 2𝑅𝑀 𝑓𝑐 − 𝑅𝑀 𝑇𝑀 𝜔𝑐 𝜔 𝜔𝑐 + 2𝜋 𝑇 𝑀 𝑓 𝐸𝐵 = 2𝑅𝑀 𝑓𝑐 + 𝑅𝑀 𝑓𝑐 + 2𝑅𝑀 114 QPSK Generation: 01101 𝑑1 𝑑3 𝑑5 𝑑7 𝑑9 𝐴 2 cos 𝜔𝑐 𝑡 + ~ 00 11 11 01 10 Polar NRZ ∑ DeMux - -π/2 𝑑1 𝑑2 𝑑3 𝑑4 𝑑5 𝑑6 𝑑7 𝑑8 𝑑9 𝑑10 0 0 1 𝑑2 𝑑4 𝑑6 𝑑8 𝑑10 1 1 01110 1 0 1 𝐴 2 1 𝜑𝑄𝑃𝑆𝐾 (𝑡) sin 𝜔𝑐 𝑡 0 I Q 𝑇𝑀 115 QPSK Demodulation 𝑥1 > 0 ⇒ 1 𝑥1 < 0 ⇒ 0 𝑥1 = 𝑎𝑖 /2 LPF Decision cos 𝜔𝑐 𝑡 𝜑𝑄𝑃𝑆𝐾 (𝑡) ~ Mux -π/2 𝑥2 = −𝑏𝑖 /2 sin 𝜔𝑐 𝑡 LPF Decision 𝑥2 < 0 ⇒ 1 𝑥2 > 0 ⇒ 0 116 Offset QPSK (OQPSK): In QPSK, the carrier phase changes only once every 𝑇𝑀 = 2𝑇𝑜 . If the two bits in a successive dibits change simultaneously (i.e. 00 then come 11 or 10 then come 01 etc.), a 180𝑜 phase shift occurs. 180𝑜 𝑠2 01 𝜓2 (𝑡) 𝑠1 11 90𝑜 𝜓1 (𝑡) 𝑠3 00 𝑠4 10 117 OQPSK (Cont.) At 180𝑜 phase-shift, the amplitude of the transmitted signal changes very rapidly costing amplitude fluctuation. This signal may be distorted when is passed through the filter or nonlinear amplifier. 118 OQPSK (Cont.) 2 1 0 -1 -2 0 1 2 3 4 5 6 7 8 Original Signal 2 1.5 1 0.5 0 -0.5 -1 -1.5 -2 0 1 2 3 4 5 6 7 8 Filtered signal 119 OQPSK (Cont.) To solve the amplitude fluctuation problem, the offset QPSK was proposed. Offset QPSK delay the data in quadrature component 𝑇𝑀 by = 𝑇𝑜 (half of symbol). 2 Now, no way that both bits can change at the same time, 𝜃𝑖 = 𝑏𝑖 −1 𝑡𝑎𝑛 𝑎𝑖 . In the offset QPSK, the phase of the signal can change by 90 or 0 degree only while in the QPSK the phase of the signal can change by 180 90 or 0 degree. 120 OQPSK (Cont.) 0 0 1 1 1 1 0 1 1 0 I Q 121 121 OQPSK (Cont.) Inphase QPSK 1 0 0.5 0 1 -0.5 0 1 -1 0 1 2 3 4 5 6 7 8 1 Q phase QPSK 0 0.5 0 -0.5 1 0 0 -1 0 1 2 3 4 5 4 5 6 7 8 2 1 QPSK 0 -1 -2 0 01 1 10 2 3 10 00 6 7 8 1 Inphase Offset QPSK 0.5 0 0 1 0 1 -0.5 -1 0 1 2 3 4 5 6 7 8 7 8 1 Q phase Offset QPSK 0.5 1 0 0 0 -0.5 -1 0 1 2 3 4 5 6 2 10 1 Offset QPSK 10 01 0 -1 00 -2 0 1 2 𝑇𝑀 3 4 5 6 7 8 122 OQPSK (Cont.) Possible paths for switching between the message points in OQPSK. 𝜓2 (𝑡) 𝑠2 𝑠1 90𝑜 𝜓1 (𝑡) 𝑠3 𝑠4 123 Hybrid Amplitude-Phase Modulation(QAM): A general form of PSK where the constraint that all points in the constellation stay on a circle is broken, this will result in QAM (Quadrature amplitude modulation). Current protocols such as 802.11b wireless Ethernet (Wi-Fi) and digital video broadcast (DVB), both utilize 64-QAM modulation. In addition, emerging wireless technologies such as Worldwide Interoperability for Microwave Access (WiMAX) and 802.11n. 124 QAM (Cont.) 𝜑𝑖 𝑡 = 𝑎𝑖 cos 𝜔𝑐 𝑡 − 𝑏𝑖 sin(𝜔𝑐 𝑡) M-ary ASK M-ary ASK 𝜆𝑇𝑀 ≤ 𝑡 ≤ 𝜆 + 1 𝑇𝑀 𝑖 = 1,2, … , 𝑀 𝜑𝑖 𝑡 = 𝑟𝑖 cos(2𝜋𝑓𝑐 𝑡 + 𝜃𝑖 ) , 𝜆𝑇𝑀 ≤ 𝑡 ≤ (𝜆 + 1)𝑇𝑀 𝑖 = 1,2, … , 𝑀 Where 𝑓𝑐 = 𝑛 𝑇𝑀 , n:integer 𝑇𝑀 = 𝑛𝑇𝑐 , 𝑇𝑐 is the period of the carrier. 125 QAM (Cont.) 𝑎𝑖 : The In-phase component (I). 𝑏𝑖 : The Quadrature component (Q). 𝑟𝑖 = 𝜃𝑖 = 𝑎𝑖 2 + 𝑏𝑖 2 𝑏𝑖 −1 𝑡𝑎𝑛 𝑎𝑖 Let 𝑧 = 𝑎𝑖 + 𝑗𝑏𝑖 called the complex envelope (the equivalent low-pass signal of 𝜑𝑖 (𝑡)). 126 M-ary PSK (Cont.) 𝜑𝑖 𝑡 = 𝑅𝑒 (𝑎𝑖 + 𝑗𝑏𝑖 )𝑒 𝑗𝜔𝑐 𝑡 = 𝑅𝑒 (𝑎𝑖 +𝑗𝑏𝑖 )(cos 𝜔𝑐 𝑡 + 𝑗 sin 𝜔𝑐 𝑡) = 𝑅𝑒 𝑎𝑖 cos 𝜔𝑐 𝑡 + 𝑗𝑎𝑖 sin 𝜔𝑐 𝑡 + 𝑗𝑏𝑖 cos 𝜔𝑐 𝑡 − 𝑏𝑖 sin 𝜔𝑐 𝑡 𝜓2 (𝑡) = 𝑎𝑖 cos 𝜔𝑐 𝑡 − 𝑏𝑖 sin 𝜔𝑐 𝑡 ⇒ 𝑧 = 𝑎𝑖 + 𝑏𝑖 = 𝑟𝑖 ⇒ ∠𝑧 = tan 𝑏𝑖 𝑎𝑖 = 𝜃𝑖 𝑏𝑖 𝑟𝑖 𝜃𝑖 𝑎𝑖 𝜓1 (𝑡) 127 𝑆𝑄𝐴𝑀 𝜔 𝜔𝑐 − 2𝜋 𝑓𝑐 − 2𝑅𝑀 𝑓𝑐 − 𝑅𝑀 𝑇𝑀 𝜔𝑐 𝜔 𝜔𝑐 + 2𝜋 𝑇 𝑀 𝑓 𝐸𝐵 = 2𝑅𝑀 𝑓𝑐 + 𝑅𝑀 𝑓𝑐 + 2𝑅𝑀 The cos(𝜔𝑐 𝑡) and sin(𝜔𝑐 𝑡) are orthogonal. PSD of QAM = PSD of 𝑎𝑖 cos 𝜔𝑐 𝑡 +PSD of 𝑏𝑖 sin(𝜔𝑐 𝑡) PSD of QAM = 2 ∙ (𝑃𝑆𝐷 𝑜𝑓 𝑀𝑎𝑟𝑦 𝐴𝑆𝐾) BW = Bandwidth of an M-ary ASK = 2𝑅𝑀 when the pulse shape is rect NRZ. 128 QAM (Cont.) General Modulator: 𝑎𝑖 cos 𝜔𝑐 𝑡 + ~ ∑ -π/2 𝑏𝑖 𝜑𝑖 (𝑡) - sin 𝜔𝑐 𝑡 129 QAM (Cont.) General Demodulator: LPF 𝑎𝑖 LPF 𝑏𝑖 2 cos 𝜔𝑐 𝑡 𝜑𝑖 (𝑡) ~ π/2 −2 sin 𝜔𝑐 𝑡 130 QAM (Cont.) 𝑘 = 𝑙𝑜𝑔2 𝑀 𝑒𝑣𝑒𝑛: 𝑠𝑞𝑢𝑎𝑟𝑒 𝑐𝑜𝑛𝑠𝑡𝑒𝑙𝑙𝑎𝑡𝑖𝑜𝑛 𝑜𝑑𝑑: 𝑐𝑟𝑜𝑠𝑠 𝑐𝑜𝑛𝑠𝑡𝑒𝑙𝑙𝑎𝑡𝑖𝑜𝑛 e.x. 𝑀 = 8 ⇒ 𝑘 = 3 ⇒ 𝑐𝑟𝑜𝑠𝑠 𝑐𝑜𝑛𝑠𝑡𝑒𝑙𝑙𝑎𝑡𝑖𝑜𝑛 𝑀 = 16 ⇒ 𝑘 = 4 ⇒ 𝑠𝑞𝑢𝑎𝑟𝑒 𝑐𝑜𝑛𝑠𝑡𝑒𝑙𝑙𝑎𝑡𝑖𝑜𝑛 𝑘⇒ 131 QAM Square constellation: 𝑘: even Let 𝐿 = + 𝑀 M-ary QAM constellation results from the Cartesian product of one dimensional L-ary ASK constellation by itself. 1-dim X 1-dim = 2-dim 2-dim X 1-dim = 3-dim 2-dim X 2-dim = 4-dim 132 QAM (cont.) e.g.: 4-ary ASK 00 01 11 -3A +A -A 10 +3A 𝜓1 (𝑡) 𝑀 = 16, 𝐿 = 4 𝜓1 𝑡 𝑋 𝜓1 𝑡 ⇒ 2-dim (-3A ,+3A), (-3A ,+A), (-3A ,-A), (-3A ,-3A), (-A ,+3A), (-A , +A), (-A , -A), (-A , -3A), (+A , +3A), (+A , +A ), ( +A , -A ), (+A , -3A), (+3A , +3A) (+3A , +A) (+3A , -A) (+3A , -3A) 133 QAM (Cont.) −3,3 (−3,1) ⇒ 𝑎𝑖 , 𝑏𝑖 = 𝐴 (−3, −1) (−3, −3) In general for any L: 1,3 (1,1) (1, −1) 1, −3 (3,3) (3,1) (3, −1) (3, −3) (−𝐿 + 1, 𝐿 − 1) (−𝐿 + 3, 𝐿 − 1) …… (𝐿 − 1, 𝐿 − 1) (−𝐿 + 1, 𝐿 − 3) (−𝐿 + 3, 𝐿 − 3) …… (𝐿 − 1, 𝐿 − 3) ⋮ ⋮ ⋮ ⋮ (−𝐿 + 1, −𝐿 + 1) (−𝐿 + 3, −𝐿 + 1) …… (𝐿 − 1, −𝐿 + 1) 𝑎𝑖 , 𝑏𝑖 = 𝐴 −1,3 (−1,1) (−1, −1) (−1, −3) 134 QAM (Cont.) 0010 0011 𝜓2 (𝑡) +3A 0110 1010 1110 +A 0111 1111 𝑟𝑖 1011 𝑏𝑖 𝜃𝑖 -3A -A +A +3A 𝑎𝑖 𝜓1 (𝑡) -A 0001 0101 0000 0100 -3A 1101 1001 1100 1000 135 QAM (Cont.) Odd – numbered dibit 𝒂𝒊 Even – numbered dibit 𝒃𝒊 00 −3𝐴 00 −3𝐴 01 −𝐴 01 −𝐴 11 +𝐴 11 +𝐴 10 +3𝐴 10 +3𝐴 4-ary ASK m> 1 4-ary ASK m> 1 136 +3𝐴cos(𝜔𝑐 𝑡) −𝐴cos(𝜔𝑐 𝑡) −3𝐴cos(𝜔𝑐 𝑡) QAM (Cont.) Input sequence 101101010010 +𝐴sin(𝜔𝑐 𝑡) 𝑘 = 4, 𝐿 = 4 −𝐴sin(𝜔𝑐 𝑡) +3𝐴sin(𝜔𝑐 𝑡) 𝐴 2cos(𝜔𝑐 𝑡 + 225𝑜 ) 𝐴 10cos(𝜔𝑐 𝑡 + 18.4𝑜 ) 𝐴 18cos(𝜔𝑐 𝑡 + 135𝑜 ) 137 QAM Modulator: 100100 𝑑1 𝑑2 𝑑5 𝑑6 𝑑9 𝑑10 L-ary PAM cos 𝜔𝑐 𝑡 + ~ 𝑑1 𝑑2 𝑑3 𝑑4 𝑑5 𝑑6 𝑑7 𝑑8 𝑑9 𝑑10 𝑑11 𝑑12 ∑ DeMux 101101010010 -π/2 - 𝜑𝑄𝐴𝑀 (𝑡) sin 𝜔𝑐 𝑡 L-ary PAM 𝑑3 𝑑4 𝑑7 𝑑8 𝑑11 𝑑12 110110 138 QAM Demodulator: 𝑥1 = 𝑎𝑖 /2 LPF Decision cos 𝜔𝑐 𝑡 𝜑𝑄𝐴𝑀 (𝑡) ~ Mux -π/2 𝑥2 = −𝑏𝑖 /2 sin 𝜔𝑐 𝑡 LPF Decision 139 Cross Constellation: 𝑘: odd 2 𝑘−1 + 4 ∗ 2𝑘−2 = 2𝑘−1 + 22 ∗ 2𝑘−3 = 2𝑘−3 𝑝𝑜𝑖𝑛𝑡𝑠 2𝑘−3 points Square 2𝑘−1 𝑝𝑜𝑖𝑛𝑡𝑠 2𝑘−1 + 2𝑘−1 = 2 ∗ 2𝑘−1 = 2𝑘 = 𝑀 points 2𝑘−3 points 2𝑘−3 𝑝𝑜𝑖𝑛𝑡𝑠 140 Cross Constellation (Cont.) 𝑀=8⇒𝑘=3 𝜓2 𝜓1 141 Cross Constellation (Cont.) 𝜓2 𝜓1 𝑀 = 32 ⇒ 𝑘 = 5 25−1 = 24 = 16, 25−3 = 22 = 4 142