Context Aware Techniques for Energy Efficient Data Acquisition in Wireless IoT for Disaster Monitoring

Abstract

Natural disasters are most dangerous events occurring in earth due to the geological,meteorological and hydrological processes. Since the natural disasters cannot be pre-vented, the only way to decrease the loss due to such incidents is to develop systems that perform continuous monitoring and gives early warning. Landslides are one of the most destructive natural disasters. Real-time wireless sensor networks is an emerging technology for continuous environmental monitoring. But, the real-world deployments are usually constrained by resources such as power, memory, and processing capabilities.We have deployed a wireless sensor network for detecting rain-fall induced landslides. The system is deployed in Munnar, Kerala, India and is operational since 2009. The system has successfully issued several warnings regarding the potential landslides in the deployment region. Most of the current research work focus to reduce communication energy to reduce the power consumption and extend the lifetime of the system. But, in most of the real-world deployments, sensing energy consumption is much higher compared to the communication and processing energy. In our landslide system, the sensing subsystem consumes 77.5%, the communication subsystem consumes 22%, and the processing subsystem consumes 0.45% of total power consumption. In this research work, we discuss a set of techniques to maximize the lifetime of the above-mentioned landslide detection system. In this system, inorder to maximize the lifetime of the system, the sensing subsystem power consumption has to be reduced. The major challenge to address is the development of techniques that reduce the power consumption while preserving the reliability of data collection and decision support by the system. This research work proposes a wavelet based sampling algorithm for choosing the minimum sampling rate for ensuring the data reliability. The results from the wavelet sampling algorithm along with the domain knowledge have been used to develop context aware data...