IODnet: Indoor/Outdoor Telecommunication Signal Detection through Deep Neural Network

In many Internet of Things (IoT) applications, understanding the device’s location can be highly significant for various reasons, including asset tracking and inventory management, geolocation services, safety and security, environmental monitoring, and proximity-based interactions. Mobile users frequently engage with mobile services/applications in both indoor and outdoor environments. Operators and other service providers can offer more suitable services to users by predicting their location accurately. Numerous attempts have been made to categorize user locations, but this paper introduces a methodology that enhances the accuracy of predicted values through Deep Neural Networks. Based on the proposed method, the accuracy of operator-side labels can be enhanced by comparing operator-side labeled datasets with real-world labeled drive-test collected datasets. The objective was to develop a precise model that rectifies uncertain labels in a dataset with more accurate labels, based on three datasets collected through crowd-sourcing and drive-testing approaches. Furthermore, the proposed method was benchmarked against state-of-the-art learning algorithms to demonstrate its superiority. Experimental results indicate that the F1-score metric can reach as high as 98% in certain datasets.