Github Repo: https://github.com/MoShamy/Body-Composition

Presentations

Knowing the composition of your body is essential to understanding your health, and achieving your fitness goals. Most people's home scales do nothing more than telling you your weight, were trying to take this to the next level. Our Body Composition Analyzer seeks to give users a more complete understanding of their body, by determining their Body Fat percentage, Total Body Water, Skeletal Muscle Mass, and other vital metrics for physical health.

Based on the ESP32 Microcontroller, our system will utilise an AC current modulator and sensor (combined with an instrumentation amplifier) to measure bodily impedence, a load cell (and subsequent ADC) to measure weight, and an ultrasonic sensor to measure height. Via a mobile app, using bluetooth, the user will be able to input their age and sex, and will be displayed the derived metrics, obtained from the measured sensory data points. The program will be running on the ESP32's built in RTOS: FreeRTOS.

To initiate the process, the user will enter the required information into the laptop, then stand on the scale. Metal electrodes will then be attached to their hands, and the ultrasonic sensor at a distance above. Once measurments are taken, and metrics generated, the GUI will present the values to the user.

• Calibrated weight acquisition subsystem utilizing a load cell interfaced through an ADC, ensuring stable and repeatable mass measurements

• Bioelectrical impedance measurement module based on controlled AC excitation and differential voltage sensing via an instrumentation amplifier, enabling extraction of raw impedance values

• Host-assisted user input interface (via laptop) for acquisition of static parameters (age, sex, and manually entered height), reducing on-device sensing complexity

• Embedded computation of Body Fat Percentage using established empirical models, integrating sensor data with user-provided parameters

• Real-time data handling and communication implemented on the ESP32 using FreeRTOS, with task-level separation for sensing, processing, and UART-based data transmission to a host display interface

• Automated height estimation subsystem using an ultrasonic sensor, enabling full on-device anthropometric data acquisition

• Extended body composition analysis including metrics such as Total Body Water, Skeletal Muscle Mass, and BMI through enhanced modeling

• Advanced user interface and data management, including a graphical dashboard and potential logging of historical measurements for trend analysis

Circuit diagrams, pinout tables, and breadboard layouts. BIA Circuit:

Estimated Total ≃ 2,000 EGP

Peak Current required per part:

Considering possible fluctuations, it would be sensible to round up to 300mA.

At an input of 5v, this would require 1.5 Watts of instantaneous power.

This project runs on FreeRTOS (native to ESP).

Task structure:

• app_main() initializes the sensors, BLE stack, and measurement manager
• ble_service_init() brings up NimBLE, configures the GATT service, and starts advertising
• ble_host_task() runs the NimBLE event loop
• measurement_manager waits for a BLE start command and runs the weight, height, and BIA cycle

BLE control plane:

• Device name: BodyComp
• Primary service UUID: 12345678-1234-5678-1234-567812345678
• Characteristics:
  • User Profile: read/write age and sex
  • Measurement Start: write 0x01 to trigger a measurement
  • Status: read/notify state, error code, and progress
  • Result: read/notify weight, height, impedance, body fat percentage, and FFM

The core measurement algorithm for bioelectrical impedance analysis. It extracts the amplitude of a specific frequency component (50 kHz) from digitally sampled ADC data with minimal computational overhead.

Purpose: Isolate the 50 kHz AC current response signal from the body impedance measurement, rejecting noise at other frequencies.

How it works:

• Uses a recursive feedback resonator with three state variables (q0, q1, q2)
• For each ADC sample, applies: q0 = coeff·q1 - q2 + x[i], then updates q2 and q1
• After processing all samples, computes magnitude: amplitude = √(real² + imag²) where:
  • real = q1 - q2·cos(ω)
  • imag = q2·sin(ω)
  • ω = 2π·f_target·n / f_sample

Why we used it in our Embedded System:

• O(n) complexity instead of O(n log n) for FFT
• Single-frequency focus eliminates unnecessary computation
• Minimal memory footprint ideal for ESP32 constraints
• Real-time capable with FreeRTOS task timing

Implementation in project:

• Processes 1024 ADC samples at 200 kHz sample rate (5.12 ms window)
• Measures impedance at injection frequency (50 kHz)
• Output amplitude directly converts to body impedance via Ohm's law and AD620 instrumentation amplifier gain

Deurenberg Single-Frequency BIA Model:

FFM (kg) = -12.44 + 0.34·(H²/Z) + 0.1534·H + 0.273·W - 0.127·A + 4.56·S

Where:

• H = height (cm)
• Z = impedance (Ω)
• W = weight (kg)
• A = age (years)
• S = sex (1=male, 0=female)

Body Fat Percentage:

BF% = (W - FFM) / W × 100

Skeletal Muscle Mass (SMM):

SMM (kg) = ((height² / R) × 0.401) + (sex × 3.825) - (age × 0.071) + 5.102

Total Body Water (TBW) — Watson Formula:

Male:   TBW = 2.447 - (0.09516 × age) + (0.1074 × height) + (0.3362 × weight)

Female: TBW = -2.097 + (0.1069 × height) + (0.2466 × weight)

Watson PE, Watson ID, Batt RD. Total body water volumes for adult males and females estimated from simple anthropometric measurements. American Journal of Clinical Nutrition, 1980.

Basal Metabolic Rate (BMR) — Mifflin-St Jeor Equation:

Male:   BMR = (10 × weight) + (6.25 × height) - (5 × age) + 5

Female: BMR = (10 × weight) + (6.25 × height) - (5 × age) - 161

Janssen I, Heymsfield SB, Baumgartner RN, Ross R. Estimation of skeletal muscle mass by bioelectrical impedance analysis. Journal of Applied Physiology, 2000.

Used for converting raw impedance measurement into clinically relevant body composition metrics.

ESP-IDF 6.1.0, ESP32 toolchain, FreeRTOS, and NimBLE BLE stack. Development and validation were done from VS Code with the ESP-IDF extension, and iPhone-side testing used nRF Connect as the BLE client. Mobile app development using Flutter & Dart.

How individual hardware components and software functions were tested in isolation.

• Connected load cell to HX711 and HX711 to ESP32 (DT → D19, SCK → D22)
• Placed known weights (1kg, 2kg) on the load cell
• Verified raw ADC counts scaled linearly with weight
• Applied tare() function to confirm zero offset on empty scale
• Pass criteria: reading within ±0.5kg of actual weight after calibration

• Connected HC-SR04 to ESP32 (TRIG → D18, ECHO → D21)
• Held a flat surface at known distances (30cm, 50cm, 100cm, 150cm, 170cm)
• Verified distance formula: distance = (duration × 0.034) / 2
• Pass criteria: reading within ±1cm of actual distance

• Tested DAC output on GPIO25 with oscilloscope
• Tested AD620AN amplifier in isolation using a known resistor (1kΩ) instead of body
• Replaced body with known resistors (200Ω, 500Ω, 1kΩ) and confirmed impedance calculation matched
• Pass criteria: impedance reading within ±10% of known resistor value

• Measured voltage at the R4/R5 divider node with a multimeter
• Confirmed stable 2.5V ± 0.1V under load
• Verified AD620 REF pin (pin 5) sits at 2.5V

• Processes 1024 samples at 200 kHz = 5.12 ms window
• Verified amplitude extraction at 50 kHz matches expected value
• Confirmed rejection of off-frequency noise components
• Pass criteria: amplitude error < 5% on clean signal, stable under 10% added noise

• Confirmed device advertises as BodyComp via nRF Connect
• Wrote age/sex to User Profile characteristic, verified readback
• Triggered measurement via 0x01 write, confirmed task unblocked
• Verified Result characteristic contains all expected metrics
• Pass criteria: all characteristics R/W, notifications received

How the system was tested as a whole.

• Connected iPhone to BodyComp via nRF Connect
• Wrote age and sex to User Profile characteristic and verified readback
• Pass criteria: profile characteristic readable/writable with correct values

• Stood on load cell with electrodes attached, triggered via BLE 0x01
• Verified both HX711 and BIA task ran concurrently without interference
• Pass criteria: both readings present in Result characteristic

• Triggered full cycle with user under HC-SR04, on scale, with electrodes
• Verified measurement manager correctly sequenced all three tasks
• Pass criteria: weight (kg), height (cm), impedance (Ω) all non-zero

• Fed real sensor readings into Deurenberg model with user profile
• Cross-checked BF% against an online BIA reference calculator
• Pass criteria: BF% within reasonable range for test subject

• Triggered full cycle from iPhone and monitored Status notifications
• Verified Result characteristic auto-updated via BLE notify on completion
• Pass criteria: all metrics delivered via notify, connection stable

Tested with 1kg reference load. Offset corrected via calibration factor.

Mostafa: Height Measurement Bluetooth Service Weight Measurement Integration

Kareem: Body Impedence Circuit Mobile App Weight Measurment Calibration & Testing

Ahmed: Initial System Architecture GUI Design & Literature Review (Component Selection, BIA Calculations) & Ideation

gantt
    title Body Composition System — Planned vs Actual Timeline
    dateFormat YYYY-MM-DD
    axisFormat %b %d
    excludes weekends


    section Planned Timeline

    Weight Measurement Module (HX711 + Load Cell) :done, p1, 2026-04-15, 5d
    Height Measurement Module (Ultrasonic) :done, p2, 2026-04-18, 7d
    BIA Integration & Validation                  :done, p3, 2026-04-28, 8d
    RTOS System Integration                       :done, crit, p4, 2026-05-04, 6d

    BLE Communication Layer                       :done, crit, p5, 2026-05-08, 5d
    Mobile App Development                        :done, crit, p6, 2026-05-10, 7d
    App ↔ ESP32 Integration & Testing             :done, crit, p7, 2026-05-14, 5d

    Final Calibration, Polish & Documentation     :done, crit, p8, 2026-05-18, 4d

    Final Project Delivery                        :milestone, m1, 2026-05-22, 0d


    section Actual Execution Timeline

    Initial Project Setup & Repository Structure  :done, a1, 2026-04-20, 3d

    Weight Measurement Module (HX711 + Load Cell) :done, a2, 2026-05-01, 3d
    Height Measurement Module (Ultrasonic) :done, a3, 2026-05-03, 4d
    BIA Circuit Integration                       :done, crit, a4, 2026-05-11, 5d

    Measurement Manager & RTOS Coordination       :done, crit, a5, 2026-05-16, 2d
    NimBLE GATT Service Integration               :done, crit, a6, 2026-05-16, 2d

    Mobile App Development & BLE Integration      :done, crit, a7, 2026-05-17, 3d

    Final Calibration, Debugging & Polish         :done, a8, 2026-05-18, 5d
    Documentation & System Finalization           :done, a9, 2026-05-21, 2d

    Final Working System                          :milestone, crit, m2, 2026-05-22, 0d

Repo Link

Iber, D. de P. (2017). Sistema de bioimpedância para a avaliação da composição corporal [Bachelor’s thesis, Centro Universitário SOCIESC – UNISOCIESC]