Sentinel-Fall: Smart Radar Based Fall Detection System

Sentinel Fall – Fall Detection System

🚀 1.0 Motivation & Story

Fig 1) Demo Video:

Sentinel-Fall is a fall detection system designed to enhance safety in elderly care, healthcare, and emergency response environments. The inspiration for this system came from a real-life incident involving my aunt and uncle, both elderly individuals with diabetes. On separate occasions, they each fell in the bathroom and, though they were calling for help, no one could hear them. They were left stranded for a significant amount of time before being discovered, resulting in serious injuries. This highlighted the urgent need for a non-intrusive, privacy-safe solution to detect falls, particularly in elderly care homes where falls are a common and dangerous issue.

Unlike traditional fall detection systems that rely on cameras or wearables, Sentinel-Fall provides a non-intrusive, reliable solution that respects privacy. By utilizing 60GHz radar technology, it ensures accurate, contactless fall detection, eliminating the discomfort of wearables and the invasiveness of cameras. This approach not only guarantees privacy but also offers a more reliable solution, reducing the false positives commonly seen in traditional systems.

The ESP32-S3 serves as the system's brain, managing processing and control while sending Wi-Fi alerts. The nRF54L15 SoC enables efficient wireless BLE (Low Energy) communication, while the SIM800L GSM Module sends emergency cellular notifications. The DFRobot SEN0623 60GHz Radar Sensor offers 100-degree horizontal and vertical coverage, ensuring reliable fall detection over a wide area (up to 11 meters when mounted at 2 meters height). Additionally, the Panasonic EKMB110111 PIR Sensor is a very low-power device that remains in sleep mode until motion is detected. When a person enters the monitored area, the sensor activates and triggers the ESP32-S3 processor, which wakes up the system. The processor then activates the radar sensor for fall detection. Only if a fall is detected does the system send alerts via Wi-Fi, GSM, and BLE, ensuring optimal power usage and long battery life.

Power-saving features are integrated by keeping most components in sleep mode when not actively detecting. The carrier board acts as an IoT motherboard, housing the BLE, GSM, and Wi-Fi modules. The system uses a 7.4V Li-ion Battery, managed by the BQ25723 Battery Charger, which allows it to charge and power the system simultaneously—just like a phone. This NVDC charger ensures continuous operation while managing energy efficiently, making the system ideal for long-term, portable use without needing constant charging.

The device can be discreetly mounted on the wall for optimal coverage and operation, while remaining portable, making it easy to carry wherever needed. With smart LEDs providing visual feedback and a buzzer delivering audible alerts, Sentinel-Fall ensures immediate notifications when a fall is detected. Additionally, the system comes with an app that allows configuration of alerts (via Wi-Fi/GSM), input of email and phone numbers, and fine-tuning of sensor parameters.

This non-intrusive, privacy-safe, and reliable system offers real-time fall detection, enhancing safety and providing peace of mind for families and caregivers, while effectively addressing a growing need in elderly care environments.

🔀2.0 Project Flow

Fig 2) Flow CHART: {VRUSHAB/ANKIT}

🧩3.0 Design Overview

📜3.1 Schematic Design & Features:

Fig ) Block Diagram of the schematic:

The Sentinel-Fall block diagram presents a compact, modular architecture for intelligent fall detection. The ESP32-S3 serves as the central processor, managing system logic and sending alerts over Wi-Fi, while the nRF54L15 handles low-power BLE communication. For cellular-based emergency notifications, the system integrates the SIM800L GSM module. Fall and presence detection are performed using a combination of a 60GHz radar sensor and a low-power PIR sensor. When a fall is detected, the system delivers immediate visual and audible alerts through smart LEDs and a buzzer.

The system is powered by a 7.4V Li-ion battery, regulated through a battery charger and a power switch, producing 3.3V, 1.8V, and 5V rails for various modules. The design includes reverse battery protection in the schematic to safeguard against incorrect battery polarity.

There are three connectors

The I2C debug connector allows expansion via external I2C modules.

The nRF module connector supports drop-in use of the Holyiot nRF54L15 BLE 5.4 module if onboard connections are unavailable

The SWD J-Link connector enables programming and debugging of the nRF54L15.

To aid prototyping and testing, test points are strategically placed across key signal lines. The PCB also includes four mounting holes for secure enclosure integration.

This power estimation table shows the worst-case current consumption of the Sentinel-Fall system, measured in milliamps (mA) for each voltage rail (5V, 4V, and 3.3V). Components like the SIM800L, ESP32-S3, and nRF54L15 are listed with their peak current draw. Based on these values, power in mW and W has been calculated per rail. This data is also used to determine the minimum PCB trace widths during layout, ensuring safe current handling without overheating or voltage drops. The green cells represent current values, aiding in accurate power draw.

🟩 3.2 PCB Design & & Layout Considerations.

To ensure robust electrical and thermal performance, all power rail widths were calculated using Saturn PCB Toolkit. Based on worst-case current consumption, a 30 mil trace width was chosen for all major power lines. This keeps the voltage drop under 0.1V and avoids overheating, even under loads of up to 3A, using 1 oz copper. For high-current paths and rail transitions, 10–25 vias are used in parallel, with each via capable of handling approximately 1.4A, ensuring adequate current density and minimal resistance. At least 2–3 vias are used for each power transition.

📷 📷

Signal integrity and routing practices were carefully follow

• USB DP and DN traces are routed with W = 2H spacing for 50 Ω impedance, using 3W spacing for loose coupling.
• All I2C lines are routed with >2W spacing from adjacent digital signals to minimize crosstalk.
• The crystal oscillator is placed as close as possible to the MCU, maintaining >20 mil clearance from other traces.
• Decoupling capacitors are placed directly adjacent to power pins for effective high-frequency filtering.
• A continuous return path is maintained for all USB signals, with reference vias added for every 4 GPIOs to ensure signal integrit
• For mechanical clarity and easier assembly, silkscreen labels are added for all connectors, including pin numbers or shorthand signal names. This is critical for avoiding confusion during PCB assembly and ensuring accurate manual or automated component placement.

The Sentinel-Fall PCB measures 60 mm × 85.16 mm and is built on a 4-layer stackup to ensure signal integrity and robust power distribution. The layer configuration includes L1 – Top (signals/components), L2 – dedicated GND plane, L3 – power plane with ground stitching, and L4 – Bottom (routing and components). This structure allows for clean signal routing, reduced EMI, and Enhanced Signal Integrity.

☐ 3.4 Enclosure Design:

This 3D-printed enclosure is purpose-built for wall mounting and features a two-part design consisting of a top lid and bottom lid for easy access and assembly. Made from PLA plastic and printed using a Prusa MK4, it has a sturdy 2.5 mm wall thickness that offers both durability and a clean finish. Inside, two mounting brackets are included to securely house the battery. Precise cutouts are provided for the PIR sensor, status LEDs, DIP switches, and the main power switch. The PCB is fastened firmly against the housing using screws, nuts, and spacers to ensure stability. The enclosure is dimensioned to fit perfectly with a phone holder wall-mount like this model, making installation quick and secure in bathroom or elderly care environments.

</> 3.3 Firmware Architecture -

a) ESP32 Code Flowchart {VRUSHAB}

b) nRF54L15 Code Flowchart {ABHIRATH}

c) Demo Video using NRF DevKit

This demo shows the nRF54L15 Development Kit sending switch button status over Bluetooth Low Energy (BLE) to the nRF Connect mobile app.
We transmit a simple payload whenever the switch state changes and also monitor the RSSI (Received Signal Strength Indicator) to track signal strength and proximity.

This forms part of the Sentinel Fall system’s low-power communication framework using BLE.

📦 4.0 Component Selection

1) C1001 60GHz mmWave Sensor

The C1001 60GHz mmWave sensor was selected after extensive research as it stands out as one of the most advanced options currently available in the market. It offers an excellent balance of long detection range (up to 11 meters for movement and 5 meters for fall/static), wide 100° detection angle, and high accuracy in fall detection using point cloud data. Its optimized power consumption (~90–100 mA) makes it suitable for always-on applications. Most importantly, its robust fall detection algorithm provides reliable performance in real-world indoor environments, making it ideal for critical safety systems like Sentinel Fall.

2) PIR Sensor: EKMB1101111

This PIR sensor was chosen because it comes as a complete, self-contained module enclosed in a durable plastic housing. Unlike raw PIR elements, this sensor includes all essential internal circuitry such as a built-in amplifier, comparator, and signal processing components, significantly simplifying integration. It ensures high reliability, EMI resistance, and minimal external components—ideal for compact, low-power embedded systems like Sentinel Fall.

3) Microcontrollers & Communication Modules:

The ESP32-S3 MINI-1 N8 was chosen for its compact form factor, integrated PCB antenna, and improved capabilities over the earlier ESP32-S2. The nRF54L15-QFAA-R is Nordic's newest ultra-low-power BLE SoC, ideal for always-on wakeup and BLE pairing with minimal energy use. Its inclusion was also a requirement of the "Kitchen Sink Award", which mandates using this specific chip. Since it lacks Wi-Fi, the ESP32 was essential for full IoT connectivity. Lastly, the SIM800L, while a legacy GSM module, was selected for its widespread adoption, available libraries, and ease of integration—making it ideal for rapid development and reliable SMS alerts in resource-constrained environments.

4) Battery Charger: BQ25886RGER

The BQ25886RGER was selected for its NVDC buck-boost architecture and integrated PowerPath feature, enabling seamless switching between battery and external input without interrupting system operation. It is programming-free, with key parameters like charging current easily configured via external resistors. Its compact design and dual power management role make it ideal for space-constrained, battery-powered applications like Sentinel Fall. Note: It does not include over-discharge protection, so it is essential to use a battery with built-in BMS (Battery Management System) for safe operation.

5) Regulators: MYLSM00502ERPL + MCP1727-3302ESN

MYLSM00502ERPL was chosen because it's a compact buck converter with an internal inductor, simplifying layout and reducing external components. The MCP1727-3302ESN is a fixed 3.3V low-dropout regulator, selected for its stability, low quiescent current, and ease of use in powering low-voltage digital circuits.

🔧💻🔨 5.0 Assem bly, Programming & Testing prcodeure:

⚙️Step 1: PCB Fabrication & Assembly with PCBWay:

To initiate fabrication, upload the following files to PCBWay:

Gerber files, Assembly file ,BOM (Bill of Materials) and Pick-and-Place file All required files are provided at the end of this article. Next, select the PCB settings exactly as shown in the provided reference images.

Place the order for both PCB manufacturing and PCB assembly together through PCBWay.

⚙️Step 2 : Firmware Programming – ESP32 & nRF54L15

• ESP32 Programming: {VRUSHAB}
  Insert the USB cable into the USB titled “ pgm usb”. Hold the BOOT switch and flash the firmware (provided at end) using the Arduino IDE. Fig
• nRF54L15 Programming: {{ABHIRATH}}
  Connect a Segger J-Link programmer to the SWD debugging connector.
  Flash the required BLE and radar control firmware using the Segger programming tools. Fig

⚙️Step 3 : Hardware Integration & Enclosure Assembly:

• 3D print the enclosure:
  Use a 3D printer (e.g., Prusa MK4) to fabricate the enclosure parts based on the provided STL or STEP files. Fig
• Secure the battery:
  Place the mounting brackets, insert the battery, and screw them in with M2.5 flat head screws, tightening them with hex nuts. Fig
• Mount the PCB:
  First, place spacers, insert the PCB into the mounting holes, and fix it using nuts. On the other side, secure the PCB tightly to the case using M3 flat head screws. Finally, connect the battery to the JST 2P connector on the board. Fig
• Refer to the provided Assembly PDF:
  Follow the step-by-step illustrated guide in the Assembly PDF to verify correct placement of all modules (GSM, PIR, Radar) and ensure full mechanical integration. Fig

⚙️Step 4 : Mounting on phone holder:

The fully assembled device was mounted onto an adjustable phone holder for final deployment. The mount’s tilt must be kep hortzonatlly facing downward, at 2-3m height, preferebly in centre of room. Else , slighly tilt and keep it at an angle. Fig

⚙️Step 5: DIP Switch Defaults:

Default DIP switch positions were configured to set the system’s boot mode and debugging options. Specific switches enabled debug UART output, active mode transition, and firmware selection where required. Proper DIP configuration ensured smooth startup and testing. Fig

⚙️Step 6 : BLE App Setup: { ABHIRATH}

The BLE mobile application was initialized by calibrating critical sensor parameters — including sensor height, fall detection timing, and presence sensitivity. User information such as emergency phone numbers and emails was entered. Software overrides for BLE, GSM, Wi-Fi, and LED notifications were enabled to allow flexible system control during testing and demonstration. Fig

⚙️Step 7 : Connectivity Setup: { VRUSHAB}

Fig/video:

• BLE: Pair with mobile app
• GSM: Verify SMS
• Wi-Fi: Configure Whats App

⚙️Step 8 : Demo Video: { VRUSHAB}

• Video Demonstrating Fall Sensing and alerts
• Includes sending alerts via BLE, GSM, and Wi-Fi.

🌱6.0 Design Challenges & Improvements

3D Casing Revisions:
The first enclosure had dimensional errors and didn't fit the phone holder. The second fit properly but had poor visibility of DIP switch, USB, and power switch engravings. Next version improved engraving clarity. Fig

Fall Sensor Troubleshooting:
Fall detection was initially unreliable. Several adjustments were made after discussions with DFRobot support to optimize radar sensitivity and positioning. (Support forum link: )

Planned Improvements:

• BLE Controller Shift: Future versions could use the nRF54L15 as the main controller, relegating ESP32 to Wi-Fi only, or move to a Nordic Wi-Fi+BLE SoC.
• Cheaper PIR Sensor: Replace the current $10 PIR with a $3 module enclosed in a custom 3D casing.
• Grounding Isolation: Separate chassis ground from PCB ground to protect against ESD from screw contacts.
• Heat-Set Inserts: Add brass inserts in top and bottom holes for stronger, reusable screw mounting as PLA is brittle.

📁7.0 Design Files & Deliverables

• PCB Assembly and Manufacturing Files::
• Design Documentation:
• Schematics:
• BOM ( Mech+ PCB):
• 3D Files(Top, Bottom Lid ,Mounting brackets):