A Hybrid TDOA-ANN Framework for Accurate Real-Time Node Localization in Wireless Sensor Network

A prototype autonomous Wireless Sensor Network node based on a microcontroller for the prognosis and forecasting of fire accidents in multi-store buildings is presented in this thesis. The system has been evaluated for its dependability and resilience, and it reacts precisely in 300 ms. One range-based localization approach that uses the Hyperbolic based least square, Taylor series based least square, and Artificial Neural Network (ANN) methods is the TDOA based localization, which aims to achieve real-time communication of sensed data. Solving the very nonlinear systems of equations with the help of the least square estimator yields the predicted locations of the sensor nodes. The suggested procedures can only be replicated with accurate TDOA measurements. Different methods can be used to calculate TDOA: one uses the hyperbolic based method, which takes into account the difference between two TOAs; another uses the Taylor series based method, which takes into account the difference between round-trip TOA measurements between a sensor node and an anchor node; and yet another uses the TDOA measurement that results from the difference between the times of arrival of Ultrasonic pulses and RF signals transmitted from a sensor to an anchor node simultaneously. The hyperbolic technique outperforms the Taylor series by a little margin and accomplishes precise localization, according to the simulation findings. The ANN technique is both practical and simple to use.