Sensor Network Based Testbench Implementation of Landslide Early Warning System

Abstract

The paper presents an implementation of prototype based Early Warning System (EWS) to detect and provide early warning of Landslide activities. The main aim of this work is to implement the prototype with low cost sensor network. A simulation setup and a table based prototype setup was implemented to study the capability and effectiveness of the system. The setup consists of table based setup of landslide with multiple/changing angle from 30° to 90°. Multiple sensing elements including rain sensor to detect presence of rain, soil moisture sensor to detect the moisture content, temperature and humidity sensor to detect humidity in the environment and vibration sensor to detect movement in the soil was implemented. The data captured by the sensing element is transmitted to a microcontroller which sends early warning signal represented by LED, Buzzer and sending of SMS or call. Through the setup, it is found that the sensitivity of the moisture sensor is rather limited however the range of moisture content detected by the sensor is functional. The vibration sensor can be tuned to have high sensitivity to the movement of the soil however, a number of false positive outcome may be detected. The early warning system is still possible by using multiple sensing element and averaging over their outcomes. The early warning system using low cost components as prototype has been successfully demonstrated. This setup can be scaled up to a real field implementation after careful tuning.