Skip to content

FFT Applications Map

Run the FFT Applications Map MicroSim Fullscreen

You can include this MicroSim on your website using the following iframe:

1
2
<iframe src="https://dmccreary.github.io/fft-benchmarking/sims/fft-applications-map/main.html"
        height="487px" width="100%" scrolling="no"></iframe>

About This MicroSim

Voice recognition, noise cancellation, machine health monitoring, radar, software defined radio, and the Wi-Fi you are probably reading this over. All six run on an FFT, and none of them requires anything you have not already built.

That last part is the point of this map. Each card names the chapter whose technique the application leans on most directly, and in every case it is a chapter you have finished:

Domain Leans on Because
Voice Recognition Chapter 16 the frame-by-frame real-time pipeline
Noise Cancellation Chapter 7 phase, which is what the complex part of a bin carries
Machine Monitoring Chapter 15 peak detection under leakage
Radar Processing Chapter 8 correlation
Software Defined Radio Chapter 6 aliasing, used deliberately
Communication Systems Chapter 13 the inverse transform

A few of these connections are worth stating out loud because they are not obvious:

Noise cancellation is a phase problem, not a magnitude problem. Getting the amplitude of the inverted copy right and the phase wrong does not reduce the noise — it can double it. That is Chapter 7's material doing real work.

Radar is correlation. A radar looks for a known transmitted pulse inside a noisy return, which is precisely the question Chapter 8 asked about a musical note. The reason it is done with an FFT is that correlation in the frequency domain is a multiplication, and multiplication is cheap.

In OFDM, the inverse FFT is the transmitter. Not a preprocessing step for the transmitter — the modulator itself. Data goes into frequency bins, an inverse FFT comes out as a waveform, and that waveform is what is broadcast.

Each card also carries a capstone project idea, sized to be genuinely buildable on a Pico 2 with the code you already have.

How to Use

  1. Hover any card for a one-line preview of what that domain does with an FFT.
  2. Click a card for the full entry: what the application is, which chapter's technique it depends on and why, and one capstone project idea.
  3. Work through all six. Then pick the one whose chapter you found most interesting, rather than the one whose domain sounds coolest — the chapter is what you will actually be doing.
  4. Press Clear selection to return to the overview.

Lesson Plan

Grade Level: Undergraduate

Duration: 10-12 minutes

Prerequisites:

  • Chapters 6, 7, 8, 13, 15, and 16 completed
  • A working FFT on the target hardware

Learning Objective: Connect each of six real-world FFT application domains back to the specific course chapter that taught its underlying technique.

Activities:

  1. Guess the chapter (4 min). Before clicking anything, show the six card names and ask students to name the chapter each one depends on most. Collect answers, then reveal. The two most often missed are noise cancellation (students say "windowing"; it is phase) and radar (students say "peak detection"; it is correlation).
  2. Trace one connection in detail (3 min). Pick machine monitoring. Ask what goes wrong if you use a rectangular window: the running-speed peak leaks across the fault frequency and hides it. This is Chapter 15's tradeoff with money on it.
  3. Choose a capstone (4 min). Have each student pick a domain and restate its project idea as a benchmark question in the form used by the Experimental Design Anatomy simulation. The two exercises are designed to be run back to back.

Assessment: Choose one of the six domains and write the three things you would need to measure to know whether your implementation works — not whether it is fast, whether it is correct.

References