Results for

100 - 119 of 327

Martin Ryba
Last activity 2023 年 12 月 12 日

AWGN SNR calibration for band limited signals

Seeing a colleague make this mistake (one I've had to fix multiple times in other's work too) makes me want to ask the community: would you like the awgn() function/blocks to give the option for creating a SNR at the bandwidth of the signal? Your typical flow is something like this:

Create a signal, usually at some nominal upsampling factor (e.g., 4) such that it's now nicely over sampled, especially if you're using a RRC or similar pulse shaping filter.
Potentially add a frequency offset (which might make the sample frequency even higher)
Add AWGN channel model for a desired SNR
Put this into your detector/receiver model

The problem is, when someone says, "I'm detecting XYZ at foo SNR," it should not magically improve as a function of the oversample. The problem isn't that awgn() generates white noise, that's what it's supposed to do and the typical receiver has noise across the entire band. The problem is that SNR is most properly defined as the signal power over the noise power spectral density times the signal's noise equivalent bandwidth. Now I looked and there's no handy function for computing NEBW for an input signal (there's just a function for assessing analysis windows). In practice it can get a bit tricky. The occupied bandwidth or HPBW are often close enough to the NEBW, we're usually not haggling over hundredths of a dB. So, in my not so humble opinion, the "measured" flag for awgn() should give an option for bandwidth matching or at least document the behavior better in the help page. All too often I'm seeing 3-6 (or worse) dB errors because people aren't taking the signal's bandwidth into account.

Highlights

Chen Lin
Last activity 2023 年 11 月 10 日

Join Community Contest 2023: Unleash Your Creativity with MATLAB Animations!

You are invited to join our 2023 community contest – MATLAB Flipbook Mini Hack! This year’s contest revolves around creating interesting animations using MATLAB.

Whether you are a seasoned MATLAB user or just getting started, this contest offers a fantastic opportunity to showcase your skills, learn from others, and engage with the vibrant MATLAB Central community.

Timeframe

This contest runs for 4 weeks from Nov. 6th to Dec. 3rd.

How to play

Create a new animation or remix an existing one with up to 2,000 characters of code.
Simply vote on the animations you love!

Prizes

You will have opportunities to win compelling prizes, including Amazon gift cards, MathWorks T-shirts, and virtual badges. We will give out both weekly prizes and grand prizes.

Check out the gallery and vote on the animations you like.

The MATLAB Central Community team