Aditya Dua (@adidua)

I have a Ph.D. in Electrical Engineering from Stanford University, where my dissertation focused on applying optimization methods to resource scheduling problems in networking. After my Ph.D., I worked at Qualcomm for 6 years, designing signal processing algorithms for 3G/4G cellular modems, then led a team of engineers at Proteus Digital Health for 5 years, working on detecting signals from a proprietary ingestible sensor. Currently, I work on health sensing related algorithms at Apple.

Filtering Noise: The Basics (Part 1)

Aditya Dua September 17, 20222 comments

Finding signals in the presence of noise is one of the fundamental quests of the discipline of signal processing. Noise is inherently random by nature, so a probability oriented approach is needed to develop a mathematical framework for filtering (i.e. removing/suppressing) noise. This framework or discipline, formally referred to as stochastic signal processing, is often taught in graduate level engineering programs and is covered from different perspectives in excellent...

Smaller DFTs from bigger DFTs

Aditya Dua January 22, 20198 comments

Let's consider the following hypothetical situation: You have a sequence $x$ with $N/2$ points and a black box which can compute the DFT (Discrete Fourier Transform) of an $N$ point sequence. How will you use the black box to compute the $N/2$ point DFT of $x$? While the problem may appear to be a bit contrived, the answer(s) shed light on some basic yet insightful and useful properties of the DFT.

On a related note, the reverse problem of computing an $N$...

Blog editor issues

New thread started 2 years ago
I am really struggling with the blog editor! In addition to MathJax equations rendering inconsistently (or not rendering at all). Additionally, I keep hitting this...

Re: Can you find the illegal step?

Reply posted 3 years ago (08/29/2021)
As a corollary, 2 = 0.

Re: Can you find the illegal step?

Reply posted 3 years ago (08/27/2021)
If $x^2 = y^2$, taking square root on both sides gives $x = +/- y$. In this case accepting the solution $x = -y$ would not have caused a contradiction.Alternatively, $x^2...

Re: Sensor data

Reply posted 3 years ago (05/31/2021)
Your professor is absolutely correct. The key to eqn 12 is the expected value operator E. If you just generate a random z and compute zz^T, there is no guarantee...

Re: Sensor data

Reply posted 3 years ago (05/30/2021)
A sample covariance computed from 10 random samples is not going to be the identity matrix.

Re: Covariance Matrix diagonal or not?

Reply posted 3 years ago (05/27/2021)
One example is not going to verify the literature. And calling a result "wrong" doesn't make sense if the example you have constructed is itself wrong. Anyway,...

Re: Covariance Matrix diagonal or not?

Reply posted 3 years ago (05/26/2021)
I should have clarified. The N=24 suggestion was for your first case only. For the last case, try N=8.Are you starting to see a pattern here? You need to pick an...

Re: Covariance Matrix diagonal or not?

Reply posted 3 years ago (05/25/2021)

Re: Slow fading channel & SNR

Reply posted 5 years ago (03/14/2019)
The SNR is a random variable in a fading environment, so have to think of it in terms of a distribution, from which you can derive summary statistics (e.g. mean...

Re: how would I model my data?

Reply posted 5 years ago (02/02/2019)

Re: Help on Noise Correlation Matrix

Reply posted 5 years ago (01/28/2019)
While noise can certainly be uniform, was your intent to generate Gaussian noise instead? in MATLAB rand produces uniformly distributed data (between 0 and 1) while...

Re: Understanding the Comb Filter Frequency Response

Reply posted 5 years ago (01/21/2019)
It looks like you have already computed the transfer function H(z) in z-domain. To compute the frequency response, you can simply set z = e^{jw) and compute H(e^jw)....

Re: Time domain signal slikes

Reply posted 5 years ago (01/14/2019)
You said they have low periodicity, but is the phenomenon periodic? The period could give you some clues about the source (e.g. any relation to the sampling rate...

Re: raised-cosine filter paramiters.

Reply posted 5 years ago (01/07/2019)
I'm not sure what "100% bad" means, but remember that rcosdesign can generate a square root raised cosine or a normal raised cosine (the latter is obtained by convolving...

Re: Not understanding digital signal processing

Reply posted 5 years ago (01/06/2019)
Every sinusoid has the following property: sin(x) = sin(x+2pi) = sin(x+4pi) ..., or compactly, sin(x) = sin(x + 2m*pi) for any integer m. Just think of the argument...

Use this form to contact adidua

Before you can contact a member of the *Related Sites:

  • You must be logged in (register here)
  • You must confirm you email address