A Proposal for Adaptive Signal Processing

Monika Aggarwal, CARE,

IIT Delhi
Tharun Kumar Reddy Bollu, ECE,
IIT Roorkee
[email protected], +91-9810032545
[email protected] , +91- 9559753536

I. Objectives of the Virtual Lab

To provide students hands-on experience and basic understanding of key concepts of Adaptive Signal
Processing. Various adaptive algorithms e.g., LMS, RLS etc. will be studied and analyzed using various
statistical and other properties. Various applications towards prediction, signal modeling, adaptive noise
cancellation, interference canceling, equalization etc. will be studied.

II. List of experiments

1. Understand various matrix operations, matrix decompositions, factorization and related operations.

2. Study the statistical properties of the output response of a system when the input is wide sense
stationary,
a. When the system is LTI
b. When the system is Linear and input is Gaussian
c. When the LTI system is modeled as AR, MA and ARMA

3. For a wide sense stationary signal evaluate the Correlogram, Blackman–Tukey, Windowed
Periodogram, Bartlett, Welch, and Daniel spectral estimates. Study various statistical properties of these
estimates with various parameters of signal.
a. For a broadband ARMA process evaluate different spectral estimates and study various statistical
properties of these estimates with various order process.
b. For a narrowband process evaluate different spectral estimates and study various statistical
properties of these estimates with various order process

4. For a broadband, narrowband ARMA process and sum of sinusoids evaluate spectral estimates by
evaluating various parameters using appropriate methods (yule walker, mini norm, least square, MuSiC
etc.) and study various statistical properties of these estimates with various order process.

5. Estimate the signal from its noisy observation using a linear filter designed by minimizing the mean
square error (Wiener Filter). Study the statistical properties error in these estimate for various filter lengths and
different observed samples.
a. Study using noises having different distribution
b. Study using different type of signals.
c. Use the algorithm to enhance signal SNR of recorded speech signal.
d. Error Performance surface analysis

6. Estimate missing values in a signal from given measurements in noise (Prediction Problem). Employ
linear MMSE predictor to estimate the missing samples in the signal based upon past and future values.
7. Simulation and comparative analysis of LMS and RLS algorithms using simulated as well real time bio-signals.
Observe the effect of various adaptation factors like step size, forgetting factor.
A) Simulate an Autoregressive stochastic process
B) Detect non-stationarity in stochastic processes with LMS and RLS

8. Study and implementation of Kalman filter.

A) Correspondence between the initial conditions of the Kalman variables and prediction performance

B) With unforced dynamic model and noiseless state space model.

9. Simulation of adaptive noise cancellation, interference canceling, channel equalization using

appropriate adaptive filters for various scenario and compare the performance for various other variants
of these adaptive algorithms.
A) Implementation and analysis of Autoregressive Stochastic processes and Minimum Variance
Distortion-less Beamformer using LMS and Monte-Carlo Runs.
B) Simulate an adaptive prediction and equalization with LMS and RLS Algorithm.

10. A) Audio synthesis using Differential DSP library.

III. Target group of users

- UG (3rd Year/ 4th Year)

IV. Mapping of proposed lab with AICTE courses of potential labs

- Adaptive Signal Processing: ECEL6

V. Mapping of proposed lab with universities

o NIT Rourkela EE4502 : Statistical Signal Processing
o IIT Ropar EEL601 Adaptive Signal Processing.
o Vignan’s University 17ES013 Adaptive Signal Processing
o Amrita Vishwa Vidyapeetham15ECE321adaptive Signal Processing

VI. Expected timelines

Activity Time line

Demo of First 3 Experiments 3 months

Demo of 4-6 Experiments 3 months

Demo of 7-10 Experiments 3 months

Validation and Final Hosting of lab 3 month

VII. Budget
Table I. Budget for ECEL6

IIT Delhi:

Budget #
S. No. Equipment/Activity (In Rupees)
1 Laptop / Machine/Printer 50,000
2 Manpower 6,00,000
3 Consumables 1,00,000
4 Contingency 2,50,000
5 Honorarium for Lab Developer (Rs 20k per 1,50,000
experiment; Ceiling of Rs 2 Lakhs per lab)
6 Miscellaneous 80,000
TOTAL 12,30,000
Budget: IIT Roorkee:

Budget #
S. No. Equipment/Activity (In Rupees)
1 Laptop / Machine/Printer 5,00,000
2 Consumables 1,50,000
3 Contingency 50,000
4 Honorarium for Lab Development 1,00,000

TOTAL 7,60,000

Note: Institute overheads is not included in the budget

VIII. Justification of the budget requirements

(a) Details of Laptop/Machine

Laptop, Desktop and printer.

(b) Details of Manpower (number, cost per man-months etc.)

a. Total man-months required: 10months
b. No. of project staff, cost per man-months: 31,000 +24%HRA
c. Honoraria for other staff associated with the project @20,000 per experiment

(c) Details of Consumables

Some audio sensors and headphone might be required to understand the effect various
noise/interference cancellation algorithm.

(d) Details of Miscellaneous cost

a. Internal Review if required: 10,000
b. Travel if required: 50,000

IX. Student Feedback and Learning

- How will you collect feedback and use them?
There will be a section of feedback at the end of each experiment where the student can give the
 feedback.
There will be set of questions at the end of each section where the student can rate each experiment.
We will continuously be looking at these feedbacks and provide an update.

- What is the actual learning component provided by the Virtual Lab?

A complete hands-on will provide students a complete understanding of all these concepts of statistical
signals, systems and adaptive system. Visualization will help them see the signal and systems behavior under
different environment and conditions. Students will be able play with the options to get more insights about
these concepts.

- After the Virtual Lab experience, would the student be able to perform the experiment in the
real lab?
After these virtual lab experiments, students will be able to do experiment in lab and will also be equipped to
handle and realize many practical systems w.r.t. noise reduction, noise/interference cancellation, equalization,
prediction (weather, stock market etc.), signal modeling (speech) etc. After these experiments, students will
also get comfortable with machine learning tools for DSP.