ESP32-Based Iot Framework for Fall Detection and Caregiver Notification

ESP32-Based Iot Framework for Fall Detection and Caregiver Notification

Nur Alisa Ali*., Abd Shukur Jaafar., Najmiah Radiah Mohamad

Fakulti Teknologi dan Kejuruteraan Elektronik dan Komputer (FTKEK)  Universiti Teknikal Malaysia Melaka (UTeM)

DOI: https://dx.doi.org/10.47772/IJRISS.2025.909000758

Received: 04 October 2025; Accepted: 12 October 2025; Published: 29 October 2025

ABSTRACT

Falls remain one of the most significant health risks for the elderly, often resulting in physical injuries, psychological trauma, and increased healthcare costs. This research introduces an IoT-enabled wearable safety band that leverages ESP32 microcontrollers, an MPU6050 motion sensor, and a NEO-6M GPS module to provide real-time fall detection and emergency alerts. Unlike conventional systems that rely on costly infrastructure, the proposed device is lightweight, affordable, and user-friendly. Once a fall is detected, the system immediately transmits notifications, including GPS coordinates, to caregivers via the WhatsApp messaging API. Audible and visual alerts from an onboard buzzer and LED further enhance user safety. Controlled testing demonstrated a detection accuracy of 95%, with a minimal false positive rate. This work highlights the potential of scalable, low-cost wearable solutions to improve elderly independence and reduce emergency response times.

Keywords: Fall detection, elderly care, real-time alert system.

INTRODUCTION

The ageing population worldwide is increasingly vulnerable to falls, which are a major cause of both physical and psychological health issues among the elderly. Age-related decline in balance, muscle strength, and other medical complications heightens the risk of accidental falls. Such incidents often lead to serious physical trauma, including fractures and head injuries, while also causing psychological consequences such as fear of recurrence, anxiety, and social isolation. According to the World Health Organization (WHO), falls represent the second leading cause of unintentional injury-related deaths across the globe [1]. With life expectancy rising and societies witnessing a steady increase in elderly populations, there is a critical need for reliable, accessible, and cost-effective fall detection technologies [2].

Beyond the human cost, falls impose a substantial economic burden on healthcare systems due to prolonged hospitalization and long-term rehabilitation. Early detection and rapid medical intervention can significantly reduce injury severity, lower treatment costs, and improve recovery outcomes [3]. For the elderly, having confidence in an assistive monitoring system also fosters a sense of independence and encourages social participation, ultimately contributing to a better quality of life ([4],[5]).

Unfortunately, many existing fall detection technologies face barriers to widespread adoption [6]. Current systems are often prohibitively expensive, limited in availability, and unsuitable for deployment in low-resource environments. Additionally, the digital divide poses challenges, as some elderly individuals may struggle to adapt to complex monitoring devices ([7],[8]).

Even though a range of solutions exists including wearable sensors, camera-based surveillance, and ambient sensing networks these approaches frequently encounter limitations in terms of their affordability, accuracy, and accessibility [9]. Continuous monitoring using high-cost equipment also makes them impractical for everyday use ([10],[11]).

To address these challenges, this research proposes the development of a lightweight, non-intrusive, and low-cost wearable device based on Internet of Things (IoT) technology. The system is designed not only to detect falls with high reliability but also to instantly notify caregivers and emergency responders, ensuring a timely response to critical incidents. Unlike conventional systems, the proposed design emphasizes compactness, affordability, and user-friendliness. Real-time alerts are sent through mobile messaging applications, complemented by audible and visual indicators that both elderly users and their caregivers can easily understand.

The primary objective of this research is to design and implement a fall detection system that can accurately identify fall events in elderly individuals. A secondary objective is to establish an automatic alert mechanism that informs healthcare professionals or designated contacts immediately after a fall is detected. The system integrates the MPU6050 motion sensor, the Neo-6M GPS module, and dual ESP32 microcontrollers (ESP32-C3 and ESP32) to create a reliable and effective prototype. By leveraging IoT technologies, the proposed system aims to bridge existing gaps in affordability, accessibility, and effectiveness, while ultimately contributing to safer, healthier, and more independent living for elderly populations.

System Design

The proposed system employs two ESP32 microcontrollers, each serving distinct roles to ensure reliable fall detection and notification. The ESP32-C3 functions as the wearable unit, continuously acquiring motion and orientation data from the MPU6050 sensor and location data from the NEO-6M GPS module. This module also integrates a buzzer and push buttons, enabling user interaction and manual overrides when necessary.

To maintain a lightweight and compact design, the system leverages the ESP-NOW communication protocol, which allows seamless data transfer between the wearable device and the secondary microcontroller acting as the communication hub. Once the ESP32-C3 detects acceleration values exceeding a predefined fall threshold, the corresponding GPS coordinates are transmitted wirelessly to the second ESP32 module.

The receiver unit (ESP32) processes the transmitted data and provides immediate local alerts. The detected fall information, including the user’s unique ID and precise location, is displayed on a Liquid Crystal Display (LCD) while a buzzer generates audible warnings to nearby individuals. To extend the alert mechanism, the receiver also employs the WhatsApp Bot API, sending real-time notifications to caregivers or family members with the user’s identification and GPS coordinates.

This two-part modular design emphasizes compactness, low power consumption, and reliability, making the system both user-friendly and practical for elderly individuals requiring continuous safety monitoring. Fig.1 shows the block diagram of the system design.

Fig. 1 The block diagram for the proposed system design

Hardware Implementation

The hardware implementation of the proposed wearable fall detection system focuses on achieving an optimal balance between functionality, compactness, and energy efficiency. The system is divided into two hardware modules: a sender and a receiver. The sender unit is designed to be compact and lightweight for the comfort of elderly users. It incorporates the ESP32-C3 microcontroller, the MPU6050 motion sensor, the NEO-6M GPS module, and push buttons for manual input. While The receiver unit functions as a terminal, receiving transmitted data from the sender via the ESP-NOW protocol. This ensures efficient, low-latency communication without relying on a standard Wi-Fi network.

ESP32-C3 Microcontroller

The XIAO ESP32-C3 by SeeedStudio was selected due to its small size, low power consumption, and strong processing capabilities based on the Espressif ESP32-C3 chip. It also integrates an external antenna to enhance wireless signal strength for Wi-Fi, Bluetooth, and radio communication. This board is compatible with the Arduino IDE, simplifying programming and prototyping. Despite its compact form factor, it provides 11 digital I/O pins that support PWM, along with 3 analog input/output pins that can be configured as ADCs. In this system, the ESP32-C3 acts as the sender, transmitting motion and location data from the MPU6050 and NEO-6M modules to the receiver.

MPU6050 Motion Sensor

The MPU6050 is a six-axis motion sensor combining a 3-axis accelerometer and a 3-axis gyroscope. It enables detection of motion, acceleration, and orientation changes in real time.

The accelerometer measures both static acceleration (gravity) and dynamic acceleration caused by movement, shocks, or vibrations. For a stationary object, the Z-axis acceleration approximates 9.8 m/s², while the X and Y axes remain close to zero. The gyroscope captures angular velocity along the roll (X-axis), pitch (Y-axis), and yaw (Z-axis), allowing precise monitoring of rotational changes.

By fusing accelerometer and gyroscope data, the MPU6050 provides more accurate orientation estimation, which is crucial for distinguishing between normal daily activities and actual fall events.

The Neo-6M GPS module is a DPS receiver that is compatible with most microcontroller boards. It provides location data, which can be used to track the patient’s location in case of an emergency or fall, enabling quick assistance. It comes with a GPS ceramic antenna, but it can be changed into any compatible antenna that is suitable for the project. For this project, a GPS ceramic antenna is used.

Push buttons used in this system will allow manual overrides and resets. The push button was set as a pull-down pushbutton which means that it by default is in a low state. When the push button is pressed, then the pin will be in a high state. Connecting with the resistor 10kOhm.

Based on the hardware for the sender, the ESP32-C3 will collect the data on acceleration and angle changes from MPU6050, when, it meets the threshold to the system, The sender will send the data from MPU6050 and GPS module Neo-6m through ESPNOW protocols to ESP32.

The sender part includes ESP32, buzzer and the LCD. ESP32 by the Espressif system functions as the communication hub. The ESP32’s built-in Wi-Fi capabilities and GPIO pins make it ideal for IoT applications. A buzzer is used to generate audible alerts in the event of a fall that can alert the caregiver. The buzzer was connected to Pin D5 of ESP32 and I2C Liquid Crystal Display (LCD) is used for this system. The schematic circuit for the Sender part is shown in Fig. 2 and the Receiver connection is shown in Fig. 3.

Fig. 2 The connection of Sender part

Fig. 3 The connection of buzzer, LCD and ESP32 as the Receiver

Software Integration

The system’s software is developed using the Arduino IDE. Key functionalities include a fall detection algorithm, ESP-NOW Protocol and the WhatsApp notification API:

The fall detection algorithm processes raw data from the MPU6050 sensor that is connected to the ESP32-C3, calculating the acceleration and angle changes of motion using accelerometer and gyroscope readings. The formula used for the accelerometer Sudden spikes in acceleration, then sudden static in acceleration and angle changes for a minute are interpreted as falls. The system then triggers alerts, sending notifications through the ESP-NOW communication protocol to the ESP32. The acceleration is determined by calculating using accelerometer data while the angle changes are determined using gyroscope data. The formulas used for calculating acceleration and angle changes are:

For this fall test algorithm, there were 2 triggers which were Trigger 1 was set as normal activities or Activity Daily Life (ADL) and Trigger 2 set as High Acceleration which had been detected by MPU6050. Fig.4 shows the testing of both trigger functions being triggered.

Fig. 4 The system testing of the Trigger function

Espressif Systems developed ESP-NOW, a low-power wireless communication protocol designed especially for ESP32 and ESP8266 devices. This protocol allows for efficient, low-latency communication between multiple devices without requiring a regular Wi-Fi network. It uses the 2.4 GHz frequency range and runs on a peer-to-peer architecture. Additionally, broadcast and unicast communications are supported by ESP-NOW. Furthermore, it can be applied to device-to-device communication, sensor networks, and remote monitoring systems. For situations requiring real-time data transfer with minimal setup, ESP-NOW is excellent.

Fig. 5 The Sender status of sending data

Fig. 6 The Receiver status of receiving data

To enable remote communication, the system integrates the WhatsApp Notification API, which allows real-time alerts to be delivered directly to caregivers. These notifications include the user’s geographical location, obtained from the GPS module, ensuring that caregivers receive immediate and actionable information during a fall incident. This rapid notification capability facilitates prompt medical response and enhances overall safety.

For message delivery, the system utilizes CallMeBot, an unofficial API service for sending WhatsApp messages. The service operates by authenticating registered phone numbers, thereby enabling secure communication between the ESP32 microcontroller and caregivers. A key advantage of CallMeBot is its ability to allow remote interaction with ESP32 devices, enabling caregivers to monitor or control the system without requiring physical access. This feature significantly improves usability, particularly in scenarios where the caregiver is not in proximity to the elderly user.

The configuration process for CallMeBot involves a few straightforward steps:

1. Save the CallMeBot phone number in the contact list.
2. Send a WhatsApp message with the text “I allow callmebot to send me messages.”
3. Wait for a confirmation reply from CallMeBot.
4. Receive a personal API key, which is then integrated into the ESP32 system code.

Once configured, the ESP32 can automatically transmit fall alerts through WhatsApp, user ID and GPS coordinates were then completed, providing an efficient and widely accessible notification platform for the caregiver.

Fig. 7 The updated status of WhatsApp Bot using CallMeBot API

In Fig. 8, the sender prototype circuit is put into an arm strap as a wearable medium.  This wearable device can be put on the arm due to its comfort and accessibility.

Fig. 8 The prototype located on the arm to become a wearable device

RESULT AND DISCUSSION

During controlled testing, the fall detection system demonstrated an overall accuracy of 95%, with a false positive rate of 4%. The average latency from fall detection to caregiver notification was approximately 30 seconds. Tests conducted with controlled falls (backwards, forward, and sideways) showed consistent detection performance.

However, activities of daily living such as rapid sitting or lying down occasionally triggered false positives, highlighting the need for further algorithm refinement. For the analysis effectiveness of the fall detection system based on the parameters using the accuracy standard formula where the system’s ability to identify the actual falls while recognizing the false falls.

where TP is the true positive, the indicated number of falls detected, and False Positive (FP) indicates false falling events detected. True Negative (TN) indicates the falling event is not detected by the system and False Negative (FN) indicates activities of daily life (ADL) are not detected as a fall event.

Table 1 shows the testing results of the system based on three different activities of daily life (ADL).

Table I:  Testing results of three different ADL

Based on Fig. 9, the Notification WhatsApp was sent from ESP32(2) that contains the ID number of the elderly, indicating that status of fall has been detected in the person. The location of the elderly was latitude and longitude. Then, the Google map location is based on the latitude and longitude given as shown in Fig. 10.

Fig. 9 The notification from WhatsApps’s application

Fig. 10 The location in GoogleMaps from the link given in WhatsApps message

The MPU6050 sensor performed reliably in detecting sudden motion changes, while the NEO-6M GPS module provided accurate location data within a 5-meter radius.

The integration of the WhatsApp API ensured timely and accessible notifications to caregivers, offering a user-friendly communication platform. Challenges included simulating slow or gradual falls and managing power consumption during prolonged use. Future work will address these issues by incorporating additional sensors and optimizing the system’s power efficiency. These results demonstrate the system’s potential to significantly enhance elderly safety and emergency response times.

CONCLUSION AND FUTURE WORK

This research has presented the successful development of a cost-effective and reliable wearable fall detection system specifically designed to meet the needs of elderly users. By integrating advanced sensor technologies with real-time communication capabilities in a compact and lightweight form factor, the system addresses significant limitations in current elderly care solutions. The findings demonstrate that the device is capable of detecting falls with high accuracy and transmitting immediate notifications, thereby enhancing both the safety and independence of older adults.

Looking ahead, several avenues for improvement have been identified. First, sensor accuracy may be enhanced through the integration of additional components, such as barometers and magnetometers, which can better distinguish between fall events and non-fall activities. Second, power optimization techniques, including advanced sleep modes, will be implemented to extend battery life and increase system sustainability. Third, the development of a dedicated caregiver application will allow real-time monitoring, improved data visualization, and stronger user–caregiver interaction.

In addition to fall detection, the system can be expanded to incorporate health monitoring features, such as heart rate, oxygen saturation (SpO₂), and blood pressure tracking, thus transforming it into a more comprehensive elderly care solution. By addressing these future enhancements, the proposed system has the potential to evolve into an integrated platform that not only improves fall safety but also contributes to broader health monitoring, ultimately increasing its clinical impact and commercial viability.

ACKNOWLEDGMENT

The authors would like to express gratitude to the Centre for Telecommunication Research & Innovation (CeTRI), Fakulti Teknologi dan Kejuruteraan Elektronik dan Komputer (FTKEK), Universiti Teknikal Malaysia Melaka for their invaluable support and resources provided throughout this research study.

REFERENCES

1. R. Scuccato, (2021). Falls in the elderly, Recent Prog Med, vol. 109, no. 7–8, pp. 401–404.
2. C. B. Pereira and A. M. K. Kanashiro, (2022). Falls in older adults: a practical approach, Arq Neuropsiquiatr, vol. 80, pp. 1-23.
3. G. C. Ang, S. L. Low, and C. H. How, (2020). Approach to falls among the elderly in the community, Singapore Med J, vol. 61, pp. 1-10
4. R. Igual, C. Medrano, and I. Plaza, (2023). Challenges, issues and trends in fall detection systems, Biomed Eng Online, vol. 12, no. 1-13.
5. C. S, Sachin, G. Monisha, H. D. SriLakshmi, S. Muteeba. (2021). Smart Fall Detection System for Elderly People Using Arduino Uno, International Journal for Research Trends and Innovation, vol. 8, pp. 76-79.
6. Y. K. Sheng and N. Zainal. (2022). Development of Wearable Sensor-Based Fall Detection System for the Elderly using IoT,” Evolution in Electrical and Electronic Engineering, vol. 3, no. 2, pp. 69–79.
7. A. Kurniawan, A. R. Hermawan, and I. K. E. Purnama (2016) A Wearable Device for Fall Detection of Elderly People Using Tri-Dimensional Accelerometer, 2016 International Seminar on Intelligent Technology and Its Applications (ISITIA), pp. 671-674.
8. T. G. Stavropoulos, A. Papastergiou, L. Mpaltadoros, S. Nikolopoulos, and I. Kompatsiaris. (2020) IoT wearable sensors and devices in elderly care: A literature review, Sensors (Switzerland), vol. 20, no. 10, pp 1-5.
9. T. Vaiyapuri, E. L. Lydia, M. Y. Sikkandar, V. G. Diaz, I. V. Pustokhina, and D. A. Pustokhin. (2021). Internet of Things and Deep Learning Enabled Elderly Fall Detection Model for Smart Homecare, IEEE Access, vol. 9, pp 1-9.
10. A. R. Nathala, E. S. Kavali, V. Raikrindhi and S. Sandiri (2023). IoT based Fall Detection System, 2023 International Conference on Advancement in Computation & Computer Technologies (InCACCT), pp. 688-691
11. Mushtaq, R., Rafique, S., Iqbal, M. W., & Ruk, S. A. (2024) Fall Detection in Elderly People. Bulletin of Business and Economics (BBE), vol. 13, pp 1-6.