Hybrid Deep Learning-Based Localization for Energy-Efficient Wireless Sensor Networks in Indoor IoT Environments

In Indoor Internet of Things (IoT) environments, where Wireless Sensor Networks (WSNs) must operate with optimal efficiency, energy-efficient localization techniques are critical. This study presents a novel hybrid deep learningbased methodology designed to enhance the precision, energy efficiency, and robustness of localization within such environments. The proposed approach integrates multiple localization techniques, including RSSI-based trilateration, inertial navigation, ultrasound-based, and magnetic fieldbased methods, with advanced deep learning models such as Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs). These models are further refined using Dynamic Feature Fusion, Adaptive Kalman Filtering, and Energy-Aware Model Optimization techniques. The methodology is evaluated using a custom dataset that simulates an indoor IoT environment. Results demonstrate that the hybrid model outperforms traditional localization techniques in terms of accuracy, scalability, and resilience to multipath interference, while maintaining high energy efficiency. This approach addresses key challenges in indoor localization, offering a reliable solution for location-aware IoT applications in resource-constrained environments.