Vibration Data Signal Processing (nRF5340 + CMSIS-DSP)
Budget: $30 – $250 USD
Project Description: Vibration Data Signal Processing (nRF5340 + CMSIS-DSP)
This project aims to design and implement the complete mathematical signal processing chain for vibration data acquired from a sensor delivering Time Waveform (TWF) acceleration signals (in g). The system will run on the Nordic nRF5340 MCU and must be optimised using the CMSIS-DSP library.
The solution must allow users to configure several processing options and obtain different types of output data, depending on their application needs.
Platform and Tools
MCU: Nordic nRF5340 (Dual-core ARM Cortex-M33)
DSP Library: CMSIS-DSP
Development Focus: Real-time performance, modular processing logic, clean and well-documented implementation
Processing Configuration – Functional Requirements
Users must be able to configure the following parameters:
1. Output Data Type
Time Waveform (TWF)
Fast Fourier Transform (FFT)
Overall Metrics:
• Peak-to-Peak
• Peak
• RMS
• Min / Max
• Standard Deviation
• Mode / Median / Average
• Skewness / Kurtosis
2. Signal Measurement Type
Acceleration
Envelope
Velocity
Displacement
3. Signal Units (Based on Measurement Type)
Acceleration: g or m/s²
Envelope: gE (g)
Velocity: mm/s or in/s
Displacement: µm or mil (1 mil = 0.001 inch)
4. Detection Type
RMS
Peak
Peak-to-Peak
5. Windowing Function
Hanning
Flattop
Uniform (Rectangular)
6. End Frequency (Sampling Dependent)
Acceleration: 1000, 2000, 4000, 5000, 6000 Hz
Envelope: 50, 100, 200, 500, 1000, 2000 Hz
Velocity: 50, 100, 200, 500, 1000 Hz
Displacement: 2, 5, 10, 20, 50 Hz
7. Envelope (demodulation) Filter Options (Only for Envelope Measurements)
Filter 1: 5–100 Hz
Filter 2: 50–1000 Hz
Filter 3: 500–6000 Hz
8. Signal Length Options (TWF / FFT)
• 1024 points to TWF = 400 lines FFT
• 2048 points to TWF = 800 lines FFT
• 4096 points to TWF = 1600 lines FFT
• 8192 points to TWF = 3200 lines FFT
• 16384 points to TWF = 6400 lines FFT
• 32768 points to TWF = 12800 lines FFT
• 65536 points to TWF = 25600 lines FFT
Deliverables
The final delivery must include:
Fully Commented Source Code
• Modular and documented
• Every step in the signal processing chain must be clearly implemented and explained
• Optimised for the CMSIS-DSP library and nRF5340 architecture
Workflow Diagram
• A complete visual representation of the signal processing pipeline
• Built in draw.io, Lucidchart, or another tool that allows export/import into Lucidchart
Comprehensive Technical Documentation
• Clear and detailed description of each processing stage
• With practical examples, variable explanations, and expected outputs
• Includes explanation of unit conversion logic, filter implementation, and FFT configuration
Expectations
Performance efficiency (low memory + fast execution)
Accurate signal transformations and metric calculations
Easy-to-understand code structure and configurable architecture
Clean coding standards and proper versioning
This project aims to design and implement the complete mathematical signal processing chain for vibration data acquired from a sensor delivering Time Waveform (TWF) acceleration signals (in g). The system will run on the Nordic nRF5340 MCU and must be optimised using the CMSIS-DSP library.
The solution must allow users to configure several processing options and obtain different types of output data, depending on their application needs.
Platform and Tools
MCU: Nordic nRF5340 (Dual-core ARM Cortex-M33)
DSP Library: CMSIS-DSP
Development Focus: Real-time performance, modular processing logic, clean and well-documented implementation
Processing Configuration – Functional Requirements
Users must be able to configure the following parameters:
1. Output Data Type
Time Waveform (TWF)
Fast Fourier Transform (FFT)
Overall Metrics:
• Peak-to-Peak
• Peak
• RMS
• Min / Max
• Standard Deviation
• Mode / Median / Average
• Skewness / Kurtosis
2. Signal Measurement Type
Acceleration
Envelope
Velocity
Displacement
3. Signal Units (Based on Measurement Type)
Acceleration: g or m/s²
Envelope: gE (g)
Velocity: mm/s or in/s
Displacement: µm or mil (1 mil = 0.001 inch)
4. Detection Type
RMS
Peak
Peak-to-Peak
5. Windowing Function
Hanning
Flattop
Uniform (Rectangular)
6. End Frequency (Sampling Dependent)
Acceleration: 1000, 2000, 4000, 5000, 6000 Hz
Envelope: 50, 100, 200, 500, 1000, 2000 Hz
Velocity: 50, 100, 200, 500, 1000 Hz
Displacement: 2, 5, 10, 20, 50 Hz
7. Envelope (demodulation) Filter Options (Only for Envelope Measurements)
Filter 1: 5–100 Hz
Filter 2: 50–1000 Hz
Filter 3: 500–6000 Hz
8. Signal Length Options (TWF / FFT)
• 1024 points to TWF = 400 lines FFT
• 2048 points to TWF = 800 lines FFT
• 4096 points to TWF = 1600 lines FFT
• 8192 points to TWF = 3200 lines FFT
• 16384 points to TWF = 6400 lines FFT
• 32768 points to TWF = 12800 lines FFT
• 65536 points to TWF = 25600 lines FFT
Deliverables
The final delivery must include:
Fully Commented Source Code
• Modular and documented
• Every step in the signal processing chain must be clearly implemented and explained
• Optimised for the CMSIS-DSP library and nRF5340 architecture
Workflow Diagram
• A complete visual representation of the signal processing pipeline
• Built in draw.io, Lucidchart, or another tool that allows export/import into Lucidchart
Comprehensive Technical Documentation
• Clear and detailed description of each processing stage
• With practical examples, variable explanations, and expected outputs
• Includes explanation of unit conversion logic, filter implementation, and FFT configuration
Expectations
Performance efficiency (low memory + fast execution)
Accurate signal transformations and metric calculations
Easy-to-understand code structure and configurable architecture
Clean coding standards and proper versioning