Intelipole: IoT and ML-Powered Smart Pole for Advanced Home Security

Abstract

Smart home is slowly but steadily becoming a part of our daily life in today\'s world. In this context, Intelipole, an IoT and Machine Learning (ML)-powered smart home system designed for advanced home security and automatic IoT device control, has been proposed. With the rise of IoT-connected devices, the integration of ML techniques is essential for enabling real-life predictions and automating device control. To achieve this, synthetic data along with a portion of real-time sensor data is collected to train the system\'s control models. Key features used include human presence count and environmental factors such as temperature, humidity, and luminosity. The system predicts appropriate control levels for devices like lights, fans, and ACs. The architecture includes real-time sensors, IP cameras for person counting, a microprocessor, cloud storage for historical data and ML models, and manual mobile control via Wi-Fi. The development process involves data collection, preprocessing, relevant feature selection, dataset splitting, and model implementation. Model performance, assessed using accuracy, precision, recall, and F1 score, shows strong potential for realworld application. Future work includes integrating user preferences and embedded systems to improve system adaptability and efficiency.

Introduction

This work presents a comprehensive IoT-enabled smart home automation system enhanced by machine learning (ML) for intelligent, real-time control of lighting, cooling, and security functions. The system is built around the Arduino microcontroller, integrates multiple real-time sensors, IP cameras, and uses a Decision Tree ML model to automate device management based on contextual environmental and human activity data.

Key Highlights:

1. Definition and Purpose of IoT in Smart Homes

• IoT refers to the interconnection of physical devices via the internet to exchange data.

• Smart homes utilize IoT to connect more devices than the number of human occupants, enhancing automation and control.

• The proposed system uses ML to improve efficiency and intelligence in managing IoT devices.

2. Proposed System Overview

• Scope: Initially developed for a single room with future expansion potential.

• Hardware Components:

  • Sensors (temperature, humidity, LDR, gas, fire)

  • IP Cameras (for person count and facial/number plate recognition)

  • Microprocessor (Arduino), DC motors, buzzer, H-Bridge

• Communication: Connected to the cloud via Wi-Fi for data storage and real-time decision-making.

• Control Mechanism:

  • Devices (lights, fans, AC) controlled using five discrete levels.

  • Mobile app allows manual overrides, which are stored to improve future ML predictions.

3. ML-Based Control System Design

• Data Input:

  • Uses both real-time sensor data and synthetically generated data.

  • Features include: Time, Temperature, Humidity, Luminosity, Person Count, etc.

• ML Model Pipeline:

  • Data Preprocessing

  • Feature Selection (via ANOVA)

  • Dataset Splitting (Train/Test)

  • Decision Tree Classifier used for control level prediction.

4. Feature-Based Model Design

• Lighting Control Model: Driven by outside luminosity and time features.

• Cooling Control Model: Combines fan and AC control, with room temperature and humidity as key features.

• Control Levels: Range from 0–4, used to set fan speeds or AC operation modes based on sensor input and predictions.

5. Security Features Using OpenCV

• Face Recognition: Triggers gate access and alerts via Telegram bot.

• Vehicle Plate Recognition: Automatically opens the gate for verified vehicles.

• Real-Time Alerts: Owner notified via Telegram with images and verification options.

6. Existing Work Review

• Summarizes past systems using:

  • GSM, GPS, EmonCMS for energy monitoring

  • ML models like LSTM, Naïve Bayes, ANN for automation

• Highlights how the current system improves upon previous models through better integration, modular control, and cloud-based intelligence.

7. Evaluation and Results

• Performance Metrics: Accuracy, Precision, Recall, F1-score using k-fold cross-validation.

• System showed strong potential for real-world deployment, with scalability and user customization.

• Telegram-based remote control and ML feedback loop offer a hybrid of automation and human override.

Conclusion

The study of large-scale Internet of Things (IoT) networks underscores a pressing need for resilient and adaptive mechanisms to manage the ever-changing patterns of network traffic and to counter increasingly sophisticated cyberattacks targeting IoT devices. In such complex environments, malicious actors frequently alter their tactics, making it essential for security systems to operate dynamically— capable of identifying and responding to a wide range of threats, including Distributed Denial of Service (DDoS) attacks, spamming, phishing, and other forms of malicious behavior. One of the key challenges in this domain is the phenomenon of concept drift, where the statistical characteristics of network traffic evolve over time. This makes static or traditional detection models insufficient, as they fail to adapt to new and unforeseen attack strategies. To tackle these challenges, this study proposes a scalable and flexible data pipeline architecture that integrates Apache Kafka, Apache Spark Structured Streaming, and MongoDB. This architecture is specifically designed to handle the highthroughput, low-latency requirements of real-time threat detection in large-scale IoT deployments. Kafka serves as a reliable and fault-tolerant messaging system for ingesting large volumes of data, while Spark Structured Streaming enables continuous data processing and dynamic model updates. MongoDB offers a robust storage solution for both historical data and model outputs, supporting efficient retrieval and analysis. A key feature of this pipeline is its ability to detect and respond to concept drift, ensuring that the security models remain accurate and responsive as network conditions change. By enabling real-time threat classification and continuous model adaptation, the proposed system significantly enhances the resilience and intelligence of IoT network defenses.

References

[1] E. Miiro and I. Kato, \"A Machine-Learning Based IoT Smart Home System to Detect and Reduce Urban Insecurity in Uganda: A Case of Kampala Metropolitan Area,\" 2024 International Conference on Artificial Intelligence, Big Data, Computing and Data Communication Systems (icABCD), Port Louis, Mauritius, 2024, pp. 1-6, doi: 10.1109/icABCD62167.2024.10645228. [2] S. Hamrioui, P. Lorenz, J. Lloret and J. J. P. C. Rodrigues, \"Cost-Effective Strategy for IIoT Security Based on Bi-Objective Optimization,\" in IEEE Internet of Things Journal, doi: 10.1109/JIOT.2025.3525779. [3] M. M. Ishaq, P. K. Sethi, S. R. Bhat, N. S. Talwandi and K. Sarawagi, \"Cloud based IoT platforms for Home Automation System,\" 2024 15th International Conference on Computing Communication and Networking Technologies (ICCCNT), 2024, doi: 10.1109/ICCCNT61001.2024.10724249. [4] V. Tsankov, B. Evstatiev and I. Valova, \"Concept for an IoT-based Electronic System for Smart Home Automation,\" 2024 9th International Conference on Energy Efficiency and Agricultural Engineering (EE&AE), Ruse, Bulgaria, 2024, pp. 1-5, doi: 10.1109/EEAE60309.2024.10600500. [5] A. Gozuoglu, O. Ozgonenel and C. Gezegin, \"Design and Implementation of Controller Boards to Monitor and Control Home Appliances for Future Smart Homes,\" in IEEE Transactions on Industrial Informatics, vol. 20, no. 9, pp. 11458-11465, Sept. 2024, doi: 10.1109/TII.2024.3404591. [6] Q. Waseem, W. I. S. Wan Din, A. Bin Ab Rahman and K. Nisar, \"Exploring Machine Learning in IoT Smart Home Automation,\" 2023 IEEE 8th International Conference on Software Engineering and Computer Systems (ICSECS), Penang, Malaysia, 2023, pp. 252-257, doi: 10.1109/ICSECS58457.2023.10256283. [7] Y. Guenfaf and Y. Zafoune, \"An IoT ML-Based system for Energy Efficiency in smart homes,\" 2023 IEEE World AI IoT Congress (AIIoT), Seattle, WA, USA, 2023, pp. 0198-0203, doi: 10.1109/AIIoT58121.2023.10174484. [8] Y. Guenfaf and Y. Zafoune, \"An IoT ML-Based system for Energy Efficiency in smart homes,\" 2023 IEEE World AI IoT Congress (AIIoT), Seattle, WA, USA, 2023, pp. 0198-0203, doi: 10.1109/AIIoT58121.2023.10174484. [9] G. Verma, S. Pachauri, A. Kumar, D. Patel, A. Kumar and A. Pandey, \"Smart Home Automation with Smart Security System over the Cloud,\" 2023 14th International Conference on Computing Communication and Networking Technologies (ICCCNT), Delhi, India, 2023, pp. 1-7, doi: 10.1109/ICCCNT56998.2023.10306548. [10] S. I. Nader, A. Das, A. Al Mamun, P. Deb Nath and G. M. Chowdhury, \"Cost-efficient Smart Home Automation and Security System based on IoT and GSM,\" 2022 International Conference for Advancement in Technology (ICONAT), Goa, India, 2022, pp. 1-5, doi: 10.1109/ICONAT53423.2022.9726112. [11] M. K. I. Shafi, M. R. Sultan, S. M. M. Rahman and M. M. Hoque, \"IoT Based Smart Home: A Machine Learning Approach,\" 2021 24th International Conference on Computer and Information Technology (ICCIT), Dhaka, Bangladesh, 2021, pp. 1-6, doi: 10.1109/ICCIT54785.2021.9689786. [12] A. Z. M. T. Kabir, A. M. Mizan, P. K. Saha, K. M. M. R. Songlap, A. J. Ta-Sin and N. A. Chisty, \"IoT Based Smart Home Automation and Security System Using Mobile App With Assistant Robot for Developing Countries,\"Jeju, Korea (South), 2021, pp. 1-4, doi: 10.1109/ICEIC51217.2021.9369770. [13] G. Lulla, A. Kumar, G. Pole and G. Deshmukh, \"IoT based Smart Security and Surveillance System,\" 2021 International Conference on Emerging Smart Computing and Informatics (ESCI), Pune, India, 2021, pp. 385-390, doi: 10.1109/ESCI50559.2021.9396843.

Copyright

Copyright © 2025 Aishwarya Jannu, Ifa Noor, Sanjana S K, Hemalatha C, Hemanth Chandra N . This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.