Skip to content

Reproducible spectral analysis bench

Demonstration project

This repository sample is executable educational work. It is not a deployed receiver, employment achievement or hardware benchmark.

Engineering problem

Make the effect of observation length and bin alignment visible without RF hardware or a scientific computing installation.

System architecture

flowchart TD
    P[Python reference DFT] --> S[SVG comparison]
    P --> A[Numerical assertions]
    C[C++ coherent accumulator] --> A
    A --> CI[GitHub Actions checks]
    S --> D[MkDocs publication]

Technologies and implementation

Python 3.12 standard library generates tones, applies periodic windows and computes a direct DFT. C++17 implements a single-bin complex accumulator. MkDocs integrates the source, figure and analysis into one reviewable publication.

Validation and testing

Run the commands in the repository README. Python checks coherent peak amplitude, leakage and far-sidelobe suppression. The compiled C++ program exits nonzero when its expected amplitude is outside tolerance. CI builds documentation strictly after the numerical checks.

Results and benchmarks

Expected synthetic result: a unit complex tone at bin 12 yields normalized amplitude 1 within \(10^{-10}\). The half-bin case distributes energy across bins. The plotted result is generated from the checked experiment.

No runtime, power, RF sensitivity or hardware benchmark has been measured. Record hardware, compiler, timing method and raw data before adding performance claims.

Technical trade-offs

A direct DFT is easy to review but scales quadratically. Standard-library-only Python reduces setup friction but is unsuitable for production throughput. An FFT implementation and captured IQ fixtures would be natural future extensions, not current capabilities.

Repository: OWNER/REPOSITORY โ€” placeholder. Set the project source URL after creating the GitHub repository.

Read the spectral leakage article.