A Complete Internet of Things based Home Security System

Abstract

Security has been a big concern especially in the unsafe areas of urban India. Theft, invasion and fire are some of the most common concerns among individuals. The existing methods of preventing the above-mentioned concerns are flawed and/or don’t address all three problems together. Most of the existing solutions rely on surveillance using cameras. In this paper, we present a complete solution which can inform the owner about invasion, send alerts for fire hazard and authenticate a visitor. Our solution uses Raspberry Pi 3B+, MQ2 gas detection module, PIR sensor, ultrasonic sensor, Arduino uno microcontroller, 3x4 keypad and a webcam. Our system is better than existing solutions as it uses ultrasonic and PIR sensors as opposed to just PIR sensors for motion detection which can lead to false positives due to high sensitivity. Also, use of arduino to connect the smoke sensors gives us a modular approach which will enable the user to connect multiple smoke sensors to the main system. Lastly, iris recognition is used for authenticating an individual. EfficientNet-Bl with SDGR optimizer was used with 300 epochs. The model had training accuracy of 99.80% and validation accuracy of 97.66%.