𝑥(𝑡) * δ(𝑡)= 𝑥(𝑡) Wienner-Khintchine theorem 𝑇0

Analog Signal Processing 1

𝑎0 = ∫ 𝑥(𝑡)𝑑𝑡
Cheat Sheet 𝑥(𝑡) * δ(𝑡 − 𝑡0)= 𝑥(𝑡 − 𝑡0) +∞
−2𝑗π𝑓𝑡
𝑇0
0
∫ 𝑅𝑥𝑥(𝑡)𝑒 𝑑𝑡
−∞
Trigonometric formula : Correlation: 𝑇0
2 2π𝑘
+∞ 𝑎𝑘 = 𝑇0
∫ 𝑥(𝑡)𝑐𝑜𝑠⁡( 𝑇0
𝑡)𝑑𝑡
𝑆𝑖𝑛 (𝑎±𝑏) = sin 𝑠𝑖𝑛 𝑎 cos 𝑐𝑜𝑠 𝑏 ± sin 𝑠𝑖𝑛 𝑏 c +∞
𝑅𝑥𝑥(0) = 𝐸 = ∫ ψ𝑥𝑥(𝑓)𝑑𝑓
0
*
𝑅𝑥𝑦(τ) = ∫ 𝑥 (𝑡). 𝑦(𝑡 − τ)𝑑𝑡 −∞ 𝑇0
𝐶𝑜𝑠 (𝑎±𝑏) = 𝑐𝑜𝑠 𝑎 𝑐𝑜𝑠 𝑏 ∓ 𝑠𝑖𝑛𝑎 sin 𝑠𝑖𝑛 𝑏 −∞ 2 2π𝑘
𝑏𝑘 = 𝑇0
∫ 𝑥(𝑡)𝑠𝑖𝑛⁡( 𝑇0
𝑡)𝑑𝑡
+∞ 0
Common analog signals: *
𝑅𝑥𝑦(τ) = ∫ 𝑦 (𝑡). 𝑥(𝑡 + τ)𝑑𝑡 Power:
−∞ ● Parseval’s theorem:
𝑠𝑔𝑛(𝑡) = {1, 𝑡 > 0 − 1, 𝑡 < 0
𝑇/2
sgn(0)=0 * 1 2 𝑇
2
∞
2
1 𝑅𝑥𝑦(τ) = 𝑅𝑦𝑥 (− τ) 𝑃= 𝑇
∫ |𝑥(𝑡) |𝑑𝑡 𝑃=
1
∫ |𝑥(𝑡)| 𝑑𝑡 = ∑ |𝑋𝑘|
𝑢(𝑡) = {1, 𝑡 > 0 0, 𝑡 < 0 u(0)= 2 −𝑇/2 𝑇
0 𝑘=−∞
𝑡 Properties : The power of a periodic signal Fourier transform:
r(t)= ∫ 𝑢(𝑠)𝑑𝑠
2
−∞
|𝑅𝑥𝑦(τ)| ≤ 𝑅𝑥𝑥(0)𝑅𝑦𝑦(0) 𝑇0/2
+∞
rect
𝑡
(𝑇) =𝑢 𝑡+ ( 𝑇
2 ) − 𝑢(𝑡 − 𝑇
2
) 𝑃 =
1
𝑇0
∫ |𝑥(𝑡) |𝑑𝑡
−𝑇0/2
2
𝑋(𝑓) = ∫ 𝑥(𝑡)𝑒
−𝑖2π𝑓𝑡
𝑑𝑡
2
𝑡𝑟𝑖(𝑡) = {1 − |𝑡|, |𝑡| < 0 0, 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒
𝑠𝑖𝑛⁡(π𝑡)
|𝑅𝑥𝑥(τ)| ≤ 𝑅𝑥𝑥(0) −∞

Sinc(t)= π𝑡
● If 0 < 𝑃 <+ ∞ Inverse Fourier transform:
+∞ 𝑅𝑥𝑥(τ) = 𝑥(τ) * 𝑥(− τ)
δ(𝑡) = {δ(𝑡) = 0, 𝑡≠0 ∫ δ(𝑡) = 1 →𝑡ℎ𝑒 𝑠𝑖𝑔𝑛𝑎𝑙 𝑖𝑠 𝑐𝑎𝑙𝑙𝑒𝑑 𝑝𝑜𝑤𝑒𝑟 𝑠𝑖𝑔𝑛𝑎𝑙 +∞
𝑖2π𝑓𝑡
−∞ Energy: 𝑥(𝑡) = ∫ 𝑋(𝑓)𝑒 𝑑𝑓
𝑑 𝑢(𝑡) 1
→𝐸 = ∞ −∞
δ(𝑡) = 𝑑𝑡
, δ(𝑎𝑡) = |𝑡|
δ(𝑡) +∞
2 If 𝑇0→∞ 𝑎𝑛𝑑 ∆𝑓𝑘→0:
Convolution: |
𝐸 = ∫ 𝑥(𝑡) 𝑑𝑡 = 𝑅𝑥𝑥(0)| Power Spectral Density (PSD):
−∞
1 2 +∞ +∞
+∞ 𝑆𝑥𝑥(𝑓) = 𝑇0
|𝑋(𝑓) | ⎡ −𝑖2π𝑓𝑠 ⎤ 𝑖2π𝑓𝑡
𝑥(𝑡) * 𝑦(𝑡) = ∫ 𝑥(τ). 𝑦(𝑡 − τ)𝑑τ For periodic signals; the energy in one period 𝑥(𝑡) = ∫ ⎢ ∫ 𝑥(𝑠)𝑒 𝑑𝑠⎥𝑒 𝑑𝑓
⎢
−∞⎣−∞
⎥
−∞
∞
⎦
𝑇0/2
Properties: 𝐸= ∫ |𝑥(𝑡) |𝑑𝑡
2 𝑃= ∫ 𝑆𝑥𝑥(𝑓)𝑑𝑓 same properties as ● Parseval’s theorem:
−𝑇0/2 −∞
ESD +∞ +∞
𝑥(𝑡) * 𝑦(𝑡)= 𝑦(𝑡) * 𝑥(𝑡) 2 2
𝐸 = ∫ |𝑥(𝑡)| 𝑑𝑡 = ∫ |𝑋(𝑓)| 𝑑𝑓
● If 0 < 𝐸 <+ ∞ −∞ −∞
If 𝑥1(𝑡) * 𝑥2(𝑡) = 𝑦(𝑡): Fourier series:
→𝑡ℎ𝑒 𝑠𝑖𝑔𝑛𝑎𝑙 𝑖𝑠 𝑐𝑎𝑙𝑙𝑒𝑑 𝑒𝑛𝑒𝑟𝑔𝑦 𝑠𝑖𝑔𝑛𝑎𝑙 ∞ 2π𝑘𝑗 ● Duality:
𝑡
( ) (
𝑥1(𝑡) * 𝑥2 𝑡 − 𝑡0 = 𝑦 𝑡 − 𝑡0 ) 𝑥(𝑡) = ∑ 𝑋𝑘 𝑒
𝑇0

→𝑃 = 0 𝑘=−∞ 𝐹𝑇{𝑥(𝑡)} = 𝑋(𝑓) ↔𝐹𝑇{𝑋(𝑡)} = 𝑥(− 𝑓)

𝑥1(𝑡 − 𝑡0) * 𝑥2(𝑡 − 𝑡1) = 𝑦(𝑡 − 𝑡0 − 𝑡1)
Energy Spectral Density (ESD): 1
𝑇0 −2π𝑘𝑗
𝑇0
𝑡 ● Convolution:
𝑋𝑘 = 𝑇0
∫ 𝑥(𝑡)𝑒 𝑑𝑡
1 0
𝑥1(𝑎𝑡) * 𝑥2(𝑎𝑡) = |𝑎𝑡|
𝑦(𝑎𝑡) ψ𝑥𝑥(𝑓) = |𝑋(𝑓) |
2
𝐹𝑇{𝑥(𝑡). 𝑦(𝑡)} = 𝑋(𝑓) * 𝑌(𝑓)
∞ ∞
𝑑𝑥1(𝑡) 𝑑𝑥2(𝑡) 2π𝑘 𝑡
𝑑𝑦(𝑡) 𝑥(𝑡) = 𝑎0 + ∑ 𝑎𝑛 * cos 𝑐𝑜𝑠 ( 𝑡) + ∑ 𝑏𝑛 * s If 𝑧(𝑡) = 𝑥( ) 𝑍(𝑓) = |𝑎|. 𝑋(𝑎𝑓)
= * 𝑥2(𝑡) = 𝑥1(𝑡) * ψ𝑥𝑥(𝑓)≥0 and ψ𝑥𝑥(− 𝑓) = ψ𝑥𝑥(𝑓) 𝑘=1
𝑇0
𝑘=1 𝑎
𝑑𝑡 𝑑𝑡 𝑑𝑡
 𝑑𝑥(𝑡) [𝑥 * 𝑦](𝑡)↔𝑋(𝑓). 𝑌(𝑓) h(t) is stable : |𝑥(𝑡)| < 𝑐𝑠𝑡𝑒(𝑏𝑜𝑢𝑛𝑑𝑒𝑑)
If 𝑧(𝑡) = 𝑑𝑡
𝑍(𝑓) = 𝑗2π𝑓 𝑋(𝑓)
(𝑛) 𝑛 ∞
𝑛
𝑑 𝑥(𝑡) 𝑛 𝑥 (𝑡) ↔ (𝑗2π𝑓) 𝑋(𝑓) and : ∫ |ℎ(τ)|𝑑τ < ∞(impulse response)
𝑧(𝑡) = 𝑛 𝑍(𝑓) = (𝑗2π𝑓) 𝑋(𝑓)
𝑑𝑡 −∞
𝑥(− 𝑡)↔𝑋(− 𝑓)
−𝑖2π𝑓𝑡0 First order Filter:
𝐹𝑇{𝑥(𝑡 − 𝑡0)} = 𝑋(𝑓) 𝑒 −𝑗2π𝑓𝑡0
𝑥(𝑡 − 𝑡0)↔𝑋(𝑓)𝑒 𝑑𝑦(𝑡)
𝑖2π𝑓0𝑡 τ 𝑑𝑡
+ 𝑦(𝑡) = 𝑥(𝑡)
𝐹𝑇{𝑥(𝑡) 𝑒 } = 𝑋(𝑓 − 𝑓0)
𝑛 𝑛 −1 (𝑛)
(− 𝑗) 𝑡 𝑥(𝑡)↔(2π) 𝑋 (𝑓) 𝑠𝑡
Fourier transform of basic analog signals: Frequency response for 1 𝑜𝑟𝑑𝑒𝑟:
Duality:
1
1↔δ(𝑓) 𝐹𝑇{𝑥(𝑡)} = 𝑋(𝑓) ● 𝐻(ω) = 𝑗τω+1

𝑠𝑖𝑛𝑐(𝑡)↔𝑟𝑒𝑐𝑡(𝑓) 𝐹𝑇{𝑋(− 𝑡)} = 𝑥(𝑓) 𝑠𝑡

Impulse response for 1 𝑜𝑟𝑑𝑒𝑟:
2
𝑠𝑖𝑛𝑐 (𝑡)↔𝑡𝑟𝑖(𝑓) 𝐹𝑇{𝑋(𝑡)} = 𝑥(− 𝑓) −τ
𝑡
1
● ℎ(𝑡) = τ
𝑒 𝑢(𝑡)
−𝑖2π𝑓0𝑡 Hilbert transform:
𝑒 ↔δ(𝑓 − 𝑓0)
𝑠𝑡
∞ Unit step response for 1 𝑜𝑟𝑑𝑒𝑟:
^ 1 𝑓(τ)
−𝑎𝑡 1 𝑥(𝑡) = ∫ 𝑑τ
𝑢(𝑡)𝑒 ↔ 𝑎+𝑗2π𝑓
π
−∞
𝑡−τ 𝑡
−τ
● 𝑦(𝑡) = ℎ(𝑡) * 𝑢(𝑡) = [1 − 𝑒 ]𝑢(𝑡)
−𝑎𝑡 1 Causal system:
𝑡. 𝑢(𝑡)𝑒 ↔ 2 Second order Filter:
(𝑎+𝑗2π𝑓)
If x(t)=0 for t≤ 0 → 𝑦(𝑡) = 0 𝑓𝑜𝑟 𝑡 ≤ 0. 2
−𝑎|𝑡| 2𝑎 𝑑 𝑦(𝑡) 𝑑𝑦(𝑡) 2 2
𝑒 ↔ 2 2 τ 𝑑𝑡
+ 2ξω𝑛 𝑑𝑡
+ ω𝑛 𝑦(𝑡) = ω𝑛 𝑥(𝑡)
𝑎 +(2π𝑓) Time invariant sys:
𝑛𝑑
𝑐𝑜𝑠⁡(2π𝑓0𝑡)↔
1
2 (
(δ 𝑓 − 𝑓0 + δ 𝑓 + 𝑓0 )) ( ) {
y(t-𝑡0)=G 𝑥(𝑡 − 𝑡0 } frequency response for 2 𝑜𝑟𝑑𝑒𝑟:
1
●
𝑠𝑖𝑛(2π𝑓0𝑡)↔
1
2𝑗 ( )
(δ 𝑓 − 𝑓0 − δ 𝑓 + 𝑓0 ) ( ) linearity: 𝐻(ω) = ω 2
(𝑗 ω ) +2ξ(𝑗 ω )+1
ω

{ }
𝑛 𝑛

2 2 G 𝑎𝑥1(𝑡) + 𝑏𝑥2(𝑡) = 𝑎𝑦1(𝑡) + 𝑏𝑦2(𝑡)

−π𝑡 −π𝑓
𝑒 ↔𝑒 ξ > 1: over damped sys.
Memoryless sys:
𝑠𝑔𝑛(𝑡)↔
1 ξ = 1: 𝑐𝑟𝑖𝑡𝑖𝑐𝑎𝑙𝑙𝑦 𝑑𝑎𝑚𝑝𝑒𝑑 𝑠𝑦𝑠.
𝑗π𝑓
y(𝑡0) = 𝑥(𝑡0)
1 1 0 < ξ < 1: 𝑑𝑎𝑚𝑝𝑒𝑑 𝑠𝑦𝑠.
𝑢(𝑡)↔ 2
δ(𝑓) + 2𝑗π𝑓 Memoryless sys→ 𝑐𝑎𝑢𝑠𝑎𝑙 𝑠𝑦𝑠
∞ ∞

𝑛=−∞
(
∑ δ 𝑡 − 𝑛𝑇0 ↔ ) 1
𝑇0
∑ δ 𝑓−
𝑛=−∞
( 𝑛
𝑇0 ) Invertible sys:

h(t)*ℎ1(𝑡) = δ(𝑡)
ⅆ
ⅆ𝑡
δ(𝑡)↔2𝑗π𝑓
Stability:
Properties of Fourier transform: