Network Capacitance Designing Synthesis
Uploaded by
ashish_soni234Network Capacitance Designing Synthesis
Uploaded by
ashish_soni234Mahindra Ecole Centrale
EE313 – Digital Signal Processing
Assignment – 03
1. Consider the following autocorrelation estimate 𝑟̂𝑥 (𝑙) ,
𝑁−𝑙−1
1
∑ 𝑥(𝑛 + 𝑙)𝑥 ∗ (𝑛) 𝑓𝑜𝑟 0 ≤ 𝑙 ≤ 𝑁 − 1
𝑟̂𝑥 (𝑙) = 𝑁
𝑛=0
𝑟̂𝑥 ∗ (−𝑙) 𝑓𝑜𝑟 − (𝑁 − 1) ≤ 𝑙 ≤ 0
{0 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒
Show that the matrix 𝑥 𝐻 𝑅̂𝑥 𝑥 is a positive definite matrix, i.e. 𝑥 𝐻 𝑅̂𝑥 𝑥 > 0 for any 𝑥 ≠ 0 .
Here, 𝑅̂𝑥 is the autocorrelation matrix formed by using the above 𝑟̂𝑥 (𝑙) .
2. Bartlett’s method is a special case of Welch’s method in which non-overlapping sections of
length L are used without windowing in the periodogram averaging operation.
(a) Show that the i-th periodogram in this method can be expressed as,
𝐿
𝑅̂𝑥,𝑖 (𝑒 𝑗𝜔 ) = ∑ 𝑟̂𝑥,𝑖 (𝑙) 𝑤𝐵 (𝑙)𝑒 −𝑗𝜔𝑙
𝑙=−𝐿
Where, 𝑤𝐵 (𝑙) is a (2𝐿 − 1) length Bartlett window.
(b) Let, 𝑢(𝑒 𝑗𝜔 ) = [1 𝑒 𝑗𝜔 … … 𝑒 𝑗(𝐿−1)𝜔 ] . Show that, 𝑅̂𝑥,𝑖 (𝑒 ) can be expressed as a
𝑇 𝑗𝜔
quadratic product,
1
𝑅̂𝑥,𝑖 (𝑒 𝑗𝜔 ) = 𝑢𝐻 (𝑒 𝑗𝜔 ) 𝑅̂𝑥,𝑖 𝑢(𝑒 𝑗𝜔 )
𝐿
Where, 𝑅̂𝑥,𝑖 is the autocorrelation matrix of 𝑟̂𝑥,𝑖 (𝑙) .
3. Consider the basic periodogram estimator 𝑅̂𝑥 (𝑒 𝑗𝜔 ) at zero frequency at 𝜔 = 0.
(a) Show that,
𝑁−1 2 𝑁−1 2
1 1
𝑅̂𝑥 (𝑒 ) = |∑ 𝑥(𝑛)𝑒 𝑗0 | = |∑ 𝑥(𝑛)|
𝑗0
𝑁 𝑁
𝑛=0 𝑛=0
(b) If 𝑥(𝑛) is a real valued, zero mean white Gaussian process with variance 𝜎𝑥2 , determine the
mean and variance of 𝑅̂𝑥 (𝑒 𝑗0 ) .
(c) Determine if 𝑅̂𝑥 (𝑒 𝑗0 ) is a consistent estimator by evaluating the variance as 𝑁 → ∞ .
4. Let 𝑣(𝑛) = 𝑥(𝑛)𝑤𝑅 (𝑛) , where 𝑤𝑅 (𝑛) is a rectangular window of length N. Show that,
1
𝑟̂𝑥 (𝑙) = 𝑣(𝑙) ⨂ 𝑣 ⋆ (−𝑙)
𝑁
Also take the DTFT of the above equation to show that,
𝑁−1
𝑅̂𝑥 (𝑒 𝑗𝜔 ) = ∑ 𝑟̂𝑥 (𝑙) 𝑒 −𝑗𝜔𝑙
𝑙=−𝑁+1
