Shiv Nadar University

Department of Electrical Engineering-(SoE)

EED364: Graph Signal Processing Lab-9 Instructor: Prof. Vijay Kumar Chakka
Topic: Signal Denoising Using Graph Spectral Filtering

I. Generating a Noise Signal:

1. Consider the given Electrocardiogram (ECG) signal () of length 5000,

having sampling frequency 500.
a. Generate and add a white Gaussian noise (()) with mean zero and
variance 0.01 to the above signal (Hint: use MATLAB
keyword)
() = () + ().
II. Signal Denoising Using Graph Spectral Filtering:

2. Denoise the signal (), by minimizing the following constraint

2
argmin {|| ||2 + }.

For minimizing, consider a 10 node graph with weighted adjacency matrix
0 1,2 1,10
1,2 0 0

1,10 0 0
2
(()())

Where , = { 22 , (() ()) , by considering = 0.1.
0,

The optimal solution of the constrained optimization problem is

1
1
() = [ ] ( ) (),
1 +
=0
Where and () are the eigenvalues and eigenvectors of above Laplacian
matrix (). ( ) is the GFT coefficients of noise signal evaluated on . Assume
= 10.
a. Plot the denoised signal ().
b. Repeat the above question with variance 0.001 and 1
c. Write your observations on the performance of the filter

Monsoon 2017 Teaching Assistants: Shaik Basheerudin Shah and Goli Srikanth