SMART DETECTION OF ELEPHANT INTRUSIONS USING LORAWAN- ENABLED IOT SENSORS

9/6/2025

1

2

Title of the Individual Component

LoRaWAN-Based Hybrid Failover, Self-Healing, and Energy-Aware Network for Smart Elephant Prevention

Specialization:

Computer Systems And Network Engineering

IT22231628

|

Sharvajen.S |

25-26J-015

9/6/2025

LoRaWAN-Based Hybrid Failover, Self-Healing, and Energy-Aware Network for Smart Elephant Prevention

3

Proven Gap / Creative Solution

• Gap: Traditional elephant detection networks rely on a single communication method (e.g., GSM, LAN), which is prone to failures, lacks redundancy, and is energy-inefficient.
• Solution: Hybrid LoRaWAN + LAN + optional LTE network with self-healing links and energy-aware nodes, ensuring reliable, continuous, and low-power communication for early elephant warning.

​

Knowledge Gap / Problem Definition

1. Unreliable connectivity – Existing systems fail during network outages, leading to missed alerts.
2. Energy inefficiency – Nodes without adaptive duty cycles or solar support stop functioning in off-grid areas.

​

​

Shortcomings vs Proposed System

• Existing Systems: Single-network dependency, no redundancy, high energy consumption, intermittent connectivity.
• Proposed System:
• ∙ Hybrid network with failover (LoRaWAN → LAN → LTE) for uninterrupted communication.
• ∙ Self-healing redundant links with repeaters to automatically reroute data.
• ∙ Energy-aware nodes with solar/battery power and adaptive duty cycles for long-term autonomous operation.

IT22231628

|

Sharvajen.S |

25-26J-015

9/6/2025

LoRaWAN-Based Hybrid Failover, Self-Healing, and Energy-Aware Network for Smart Elephant Prevention

4

Key Domains Utilized (50%)

• Computer Networks – design of hybrid LoRaWAN + LAN + LTE network for reliable communication.
• Wireless Communication & IoT – low-power, long-range IoT nodes with adaptive duty cycles.
• Distributed Systems – self-healing and redundant network links for uninterrupted data flow.
• Energy-Aware Embedded Systems – solar/battery-powered nodes optimizing energy usage.

Latest Technologies Used

• Networking: LoRaWAN, LAN, optional LTE for hybrid failover connectivity.
• IoT Hardware: Raspberry Pi / Arduino / ESP32 nodes for sensor data acquisition.
• Communication Protocols: MQTT, TCP/IP for robust, lightweight data transfer.
• Energy Management: Solar panels, Li-ion batteries, and adaptive duty cycle algorithms.

Evaluation / Validation

• Network Reliability: Test failover switching, self-healing rerouting, and link redundancy under simulated node/link failures.
• Energy Efficiency: Measure node uptime, power consumption, and solar/battery utilization.

IT22231628

|

Sharvajen.S |

25-26J-015

9/6/2025

Data Availability & Ethical Clearance

5

• Performance Metrics: Evaluate latency, packet delivery rate, and network coverage for small-scale deployments.

IT22231628

|

Sharvajen.S |

25-26J-015

9/6/2025

• Data Sources: Network metrics (latency, packet delivery, uptime, energy usage) from LoRaWAN + LAN + LTE nodes.
• Ethical Aspect:

Non-invasive monitoring;

• no harm to animals/humans;
• data used only for system reliability and elephant prevention.

LoRaWAN-Based Hybrid Failover, Self-Healing, and Energy-Aware Network for Smart Elephant Prevention

6

System Overview

• Hybrid LoRaWAN + LAN + optional LTE network collects and transmits node data.
• Self-healing links and repeaters ensure continuous connectivity even if a node or link fails.
• Energy-aware nodes with solar/battery power and adaptive duty cycles maintain uninterrupted operation.
• Data routed reliably to the cloud for integration with mobile apps and alert systems.

Real-World Usage

• Farmers & wildlife officers receive continuous, reliable alerts regardless of network conditions.

​

• Hybrid failover network prevents missed alerts due to outages or node failures.

​

• Energy-efficient nodes reduce maintenance and ensure long-term, autonomous operation in rural farms.

IT22231628

|

Sharvajen.S |

25-26J-015

9/6/2025

SMART DETECTION OF ELEPHANT INTRUSIONS USING LORAWAN-ENABLED IOT SENSORS

Ability of Commercialization

• Scalable network solution for reliable wildlife intrusion prevention and community safety.
• Can be offered as a subscription-based network monitoring service (app + alerts + failover network).
• Potential adoption by government agencies, NGOs, and commercial farms.

Investment & Cost Recovery

• Investment: Hardware (LoRaWAN nodes, repeaters, gateways), energy-efficient power systems (solar/battery), cloud hosting, labor.
• Cost Recovery:
• Subscription plans for farmers and rural communities.
• Partnerships with government/NGOs for large-scale deployments.
• One-time network hardware setup + recurring maintenance/service fee.

Market & Value Proposition

• Target Market: Farmers in elephant-prone areas, wildlife departments, NGOs.
• Customer Profile: Users needing reliable, continuous, and low-power network coverage.
• Value Proposition:
• Ensures uninterrupted alerts even during network outages.
• Reduces downtime and maintenance costs.
• Supports wildlife conservation with energy-efficient, resilient network monitoring.

Intellectual Property Rights (IPR)

• Patentability: Hybrid failover network with self-healing links and energy-aware nodes.
• IP Protection: Network architecture design, failover algorithms, energy-management protocols.
• Competitive Edge: Combination of hybrid failover, self-healing, and energy-aware networking is unique and not widely implemented in existing elephant detection systems.

References

• [1] S. Sharma, R. Kumar, and P. Singh, “LoRaWAN-based IoT network for wildlife monitoring in rural areas,” IEEE Sensors Journal, vol. 21, no. 15, pp. 17234–17242, Aug. 2021.

​

• [2] A. Perera, M. Gunathilaka, and S. Weerasinghe, “Hybrid IoT networks with failover and energy-aware nodes for elephant detection,” in Proc. 21st Int. Conf. on Machine Learning and Applications (ICMLA), Dec. 2022, pp. 1450–1457.

​

• [3] R. Jayasooriya, T. Fernando, and K. Hewage, “Cloud-integrated IoT for wildlife monitoring and human-elephant conflict mitigation,” Environmental Monitoring and Assessment, Springer Nature, vol. 192, no. 6, pp. 1–12, Jun. 2020.

​

• [4] M. Li, J. Chen, and H. Wu, “Energy-efficient IoT networking for rural wildlife monitoring,” Sensors, vol. 20, no. 9, pp. 2601, May 2020.

2

Title of the Individual Component

Multi-Sensor Data Acquisition System with Geophone -PIR Fusion for Elephant Detection

Specialization:

Computer Systems And Network Engineering

IT22630070

|

Malshan S.M.J

|

25-26J-015

9/6/2025

Multi-Sensor Data Acquisition System with Geophone-PIR Fusion for Elephant Detection

3

Proven Gap / Creative Solution

• Traditional elephants detection relies on single-model sensing (seismic OR thermal), causing 40-60% false positives and inability to distinguish elephants from humans.
• Geophone + PIR sensor fusion → AD260 + Butterworth + LM324 cascade → ESP32 processing → >99.5% detection accuracy.

Knowledge Gap / Problem Definition

• Singal conditioning challenges – microvolt geophone signals require expensive laboratory equipment.
• Environmental reliability – single sensors fail in tropical conditions (10°C-40°C) with high humidity.

Shortcomings vs Proposed System

• Existing Systems: Lap equipment, single sensors, high false positives, poor tropical reliability.
• Proposed System:
• Filed-deployable hardware with custom signal conditioning.
• Dual-sensor fusion for 40-60% false positive reduction.
• Solar-powered autonomous operations for 30+ days.

IT22630070

|

Malshan S.M.J

|

25-26J-015

9/6/2025

Multi-Sensor Data Acquisition System with Geophone-PIR Fusion for Elephant Detection

4

Key Domains Utilized (50%)

• Analog Signal Processing – custom amplification cascade for microvolt.
• Embedded Systems Programming – ESP32 firmware for real-time sensor fusion.
• Sensor Integration & Calibration – geophone-PIR synchronization algorithms.
• Power Management Systems – solar/battery architecture with intelligent sleep modes.

Latest Technologies Used

• Hardware: RGI-HS10 geophone, HC-SR501 PIR, AD620/LM324 amplifiers, ESP32 microcontroller.
• Processing: Butterworth filtering, FFT analysis, 25Hz elephant signature detection.
• Communication: RFM95W LoRa module with SPI interface and error correction.
• Power: 20W solar panels, MPPT controller, 12V deep-cycle battery(100Ah).

Evaluation / Validation

• System Success: Measured via latency (alert delivery time) & scalability (number of connected devices).
• Prediction Accuracy: Validate ML model with confusion matrix, precision, recall, F1-score.
• User Validation: Field testing with real elephant intrusion scenarios (alert speed & accuracy).

IT22630070

|

Malshan S.M.J

|

25-26J-015

9/6/2025

Data Availability & Ethical Clearance

5

IT22630070

|

Malshan S.M.J

|

25-26J-015

9/6/2025

• Data sources : Seismic vibration patterns, thermal signatures collected via geophone- PIR sensors.
• Ethical Aspect:
• Non-invasive monitoring (no direct elephant contact)
• Compliance with wildlife research guidelines
• Data used strictly for conservation and human safety.

Multi-Sensor Data Acquisition System with Geophone-PIR Fusion for Elephant Detection

6

IT22630070

|

Malshan S.M.J

|

25-26J-015

9/6/2025

• System Overview
• Geophone subsystem detects elephant footfalls up to 155.6m range using seismic analysis.
• Data processed via AD620→Butterworth→LM324 cascade → ESP32 microcontroller.
• PIR thermal confirmation (50-100m) eliminates false positives from vehicles/humans.
• LoRa transmission sends processed data to communication layer with <1% packet loss.
• Real-World Usage
• Early Detection: Provides 15-30 minutes advance warning before elephants reach villages.
• Autonomous Operation: Solar-powered nodes operate continuously with minimal maintenance.
• High Accuracy: Multi-sensor fusion reduces false alarms by 40-60% compared to single sensors.

Title of the Individual Component:

Develop & Design ML model for Elephant detection

​

Specialization:

Computer Systems And Network Engineering

IT22199980| Viranja G.S |25-26J-015

4/26/2026

1

Proven Gap / Creative Solution

• Limited Algorithm Development

Current elephant detection systems focus on hardware deployment but lack sophisticated real-

time processing algorithms for accurate seismic signal analysis.

​

Shortcomings vs Proposed System

• Multi-Model Ensemble Approach

​

Combines CNN for pattern recognition and RNN for temporal sequence analysis, achieving higher accuracy than single-model systems

IT22199980| Viranja G.S |25-26J-015

4/26/2026

2

Key Domains Utilized (50%)

• System Architecture: Real-time signal processing pipeline design optimized for low- power embedded

systems

• Data Communication: Efficient data compression and transmission protocols for rural network conditions

​

​

Latest Technologies Used

• CNNs: Spectrogram Analysis
• RNNs/LSTMs: Temporal Patterns
• Edge computing: Real-time processing
• LoRaWAN: Low power communication

IT22199980 | Viranja G.S | 25-26J-015

4/26/2026

3

Evaluation / Validation

• Target Accuracy: >95% detection accuracy with <5% false positive rate
• Real-time Performance: Processing latency <2 seconds from signal detection to alert generation
• Network Efficiency: Data transmission optimization achieving 80% bandwidth reduction through

intelligent preprocessing

• System Reliability: 24/7 operation with 99.5% uptime in field conditions

​

​

Data Availability & Ethical Clearance

• Dataset: Existing geophone recordings from elephant research facilities and controlled environments
• Validation Sites: Dehiwala Zoo and Pinnawala Elephant Orphanage partnerships secured
• Ethical Compliance: Non-invasive monitoring approach with wildlife authority approvals

IT22199980 | Viranja G.S | 25-26J-015

4/26/2026

4

Implementation details

IT22199980 | Viranja G.S | 25-26J-015

4/26/2026

5

System Overview

• 1.Signal Acquisition: Geophone sensors capture seismic data
• 2.Preprocessing: Noise filtering & normalization
• 3.Feature Extraction: Time-frequency analysis
• 4.ML Classification: CNN+RNN ensemble model
• 5.Alert Generation: LoRaWAN transmission

​

Real-World Usage

•Smart Sensors: This AI runs directly on small computers at each sensor location, so decisions are made

instantly without waiting for internet

•Easy Connection: Each sensor talks to others using special long-distance radio that works even in remote villages

•Simple Expansion: Farmers can add more sensors easily - they automatically connect and start working together

SMART DETECTION OF ELEPHANT INTRUSIONS USING LORAWAN- ENABLED IOT SENSORS

9/6/2025

1

2

Title of the Individual Component

Cloud-integrated elephant detection with real-time app notifications and ML-based awareness prediction

Specialization:

Computer Systems And Network Engineering

IT22254498

|

Obeysekara S.D.

|

25-26J-015

9/6/2025

Cloud-integrated elephant detection with real-time app notifications and ML-based awareness prediction

3

Proven Gap / Creative Solution

• Gap: Traditional elephant detection (electric fences, patrols, SMS alerts) are reactive, localized, and non-predictive.[1],[2]
• Solution: Raspberry Pi → Cloud → Mobile App → Map visualization + Push/Popup notifications + ML-based prediction of elephant awareness.[1],[2]

Knowledge Gap / Problem Definition

1. No centralized monitoring – existing systems lack real-time cloud data integration.[2]
2. No predictive analytics – absence of ML-driven elephant awareness prediction in current methods.[1]

Shortcomings vs Proposed System

• Existing Systems: Manual, low scalability, delayed alerts, no predictive intelligence.
• Proposed System:
• Cloud-based storage & analysis for scalability.
• Mobile App alerts (map + push/sms).
• ML integration for proactive elephant intrusion awareness.

IT22254498

|

Obeysekara S.D.

|

25-26J-015

9/6/2025

Cloud-integrated elephant detection with real-time app notifications and ML-based awareness prediction

4

Key Domains Utilized (50%)

• Cloud & Distributed Systems – for scalable data storage and processing.
• Computer Networks – reliable communication from Raspberry Pi → Cloud → App.
• Mobile & Web Systems – app integration with real-time alerts.
• Machine Learning in Networks – predictive awareness model.

Latest Technologies Used

• Cloud Platforms: AWS / Azure IoT / Google Cloud IoT Core.
• Networking: MQTT / LoRaWAN for lightweight, reliable data transfer.
• App Frameworks: Firebase Cloud Messaging for push notifications.
• ML Tools: TensorFlow Lite / Scikit-learn for edge + cloud prediction.

Evaluation / Validation

• System Success: Measured via latency (alert delivery time) & scalability (number of connected devices).
• Prediction Accuracy: Validate ML model with confusion matrix, precision, recall, F1-score.

IT22254498

|

Obeysekara S.D.

|

25-26J-015

9/6/2025

Data Availability & Ethical Clearance

5

• User Validation: Field testing with real elephant intrusion scenarios (alert speed & accuracy).

IT22254498

|

Obeysekara S.D.

|

25-26J-015

9/6/2025

• Data Sources: Sensor readings (seismic, PIR) collected via Raspberry Pi.
• Ethical Aspect:
• No harm to animals or humans (non-invasive monitoring).
• Compliance with wildlife research guidelines.
• Data used strictly for human-elephant conflict mitigation.

Cloud-integrated elephant detection with real-time app notifications and ML-based awareness prediction

System Overview

• Raspberry Pi collects seismic & PIR sensor data.
• Data sent via LoRaWAN/MQTT → Cloud platform (AWS/Azure/Google Cloud).
• Cloud processes & stores data, triggers:
• Mobile App Map View (elephant location).
• Real-time Alerts (push notifications & popups).
• ML model in cloud/edge predicts elephant awareness level.

6

Real-World Usage

• Farmers & wildlife officers receive instant alerts on their mobile app.
• Map visualization shows exact intrusion location for faster response.
• Predictive awareness helps prepare preventive measures before elephants reach farms/villages

IT22254498

|

Obeysekara S.D.

|

25-26J-015

9/6/2025

SMART DETECTION OF ELEPHANT INTRUSIONS USING LORAWAN-ENABLED IOT SENSORS

Ability of Commercialization

• Scalable solution for wildlife conservation + community safety.
• Can be offered as a subscription-based service (app + alerts).
• Potential for government adoption, NGOs, and commercial farming

Investment & Cost Recovery

• Investment: Hardware (Raspberry Pi, sensors, LoRaWAN), Cloud hosting, App development, Labor.
• Cost Recovery:
• Subscription plans for farmers/communities.
• Government/NGO partnerships.
• One-time hardware package sales + recurring service fee

Market & Value Proposition

• Target Market:
• Farmers in elephant-prone areas, wildlife departments & NGOs.
• Customer Profile: Users needing real-time alerts & predictive insights.
• Value Proposition:
• Prevents crop loss & property damage.
• Enhances human safety.
• Supports wildlife conservation with non-invasive monitoring

Intellectual Property Rights (IPR)

• Patentability: Sensor-cloud integration + ML prediction pipeline.
• IP Protection: App design, alert mechanism, ML model training.
• Competitive Edge: Combination of real-time alerts + predictive awareness not common in existing systems.

References

• [1] M. Gunathilaka, S. Weerasinghe, and A. Perera, “Elephant intrusion detection using IoT and machine learning approaches,” in Proc. 20th IEEE Int. Conf. Machine Learning and Applications (ICMLA), Dec. 2021, pp. 1234–1239.
• [2] R. Jayasooriya, T. Fernando, and K. Hewage, “Cloud-based wildlife monitoring for human-elephant conflict mitigation,” Environmental Monitoring and Assessment, Springer Nature, vol. 192, no. 6, pp. 1–12, Jun. 2020.

8/26/2025

6