Fall Detection System for Elderly using ESP32 with SMS Alerts

Older adults who live alone face a quiet, serious risk: a fall that leaves them unable to call for help, with assistance arriving hours later. Commercial fall-detection pendants exist but are expensive and often tied to paid subscription services, putting them out of reach for many families. This project builds an affordable wearable that watches the wearer's motion with an accelerometer-gyroscope sensor, decides on the spot whether a fall has likely happened, and — unless the wearer cancels the alarm — sends SMS alerts directly to family members' phones through a GSM module, with no subscription service involved. It is also a strong teaching build: the student works with real IMU signal processing (acceleration magnitude, orientation change, inactivity windows), embedded firmware on the ESP32, AT-command control of a GSM module, and the practical design of a small wearable device. The detection logic is threshold-based and tuned on a bench prototype, so the honest engineering trade-offs — sensitivity versus false alarms — are part of the learning, not hidden.

1. The device is worn at the waist and powered on; the MPU6050 measures acceleration and angular rate while the ESP32 sleeps between motion interrupts to save power.
2. A motion interrupt wakes the ESP32, which reads the sensor stream and computes the acceleration magnitude along with the device's orientation from the gyroscope.
3. The firmware applies the three-stage fall signature: an impact spike above the configured threshold, followed by a rapid orientation change, followed by roughly 10 seconds of stillness indicating the wearer may not have got up.
4. If the signature matches, the piezo buzzer sounds and the approximately 30-second cancel window starts (duration adjustable in code).
5. If the wearer presses the cancel button within the window, the event is discarded and the device returns to low-power monitoring with no message sent.
6. If nobody cancels, the ESP32 wakes the SIM800L, registers on the GSM network, and sends the alert SMS to each stored emergency contact in turn.
7. The event is logged with a timestamp over the USB serial port; the device then returns to low-power monitoring until the next motion event.

Tech stack:

• ESP32 (DevKit-style board)
• MPU6050 6-axis IMU
• SIM800L GSM module
• Arduino IDE · C/C++ firmware
• AT-command SMS handling
• Piezo buzzer · push-button cancel
• Li-Po battery · micro-USB charging

• [Three-stage fall check] The ESP32 samples the MPU6050 accelerometer and gyroscope continuously and flags a fall only when three conditions occur in sequence: a sharp acceleration spike, a sudden change in body orientation, and a short window of stillness afterwards. This sequencing filters out ordinary jerks and bumps.
• [SMS alerts via SIM800L] On a confirmed, uncancelled fall, the SIM800L GSM module sends an SMS to each stored emergency contact (up to 3 numbers, configurable in the code), carrying a "Fall detected" message and a configurable location note such as the wearer's home address.
• [Cancel window with local buzzer] The device sounds the piezo buzzer and gives the wearer approximately 30 seconds to press the cancel button before any SMS is sent, which is how the occasional false positive from sitting down quickly or dropping the device is handled.
• [Stored contact list] Emergency numbers are stored in the firmware configuration and can be changed with a simple edit and re-upload; no external app or subscription is required.
• [Low-power firmware design] The ESP32 is placed in deep sleep between motion events, waking on the MPU6050's motion interrupt, and the SIM800L is kept in sleep mode except when sending alerts — a design intended to stretch battery life over days of wear.
• [Waist-strap wearable enclosure] The electronics sit inside a fabric pouch mounted on an elastic waist strap at the hip, where falls produce the clearest motion signature; the strap fits typical adult waist sizes and the pouch recharges over micro-USB.
• [Event logging over serial] Each detected fall and cancel event is printed with a timestamp to the USB serial port, so the student can verify behaviour and tune thresholds during the demo and the viva.

• Working prototype: waist-strap fall-detection device in fabric pouch enclosure
• Complete ESP32 firmware source code (Arduino IDE)
• Circuit and wiring documentation with connection table
• Component list with exact part numbers and sourcing guidance
• Setup, wearing and threshold-tuning instructions
• Project abstract PDF
• Project report PDF (background, methodology, limitations, results of bench tests)
• PPT presentation for final review
• Viva Q&A preparation document (IMU concepts, firmware logic, GSM/AT commands, false-positive handling)