An Innovative IoT Framework using Machine Learning for Predicting Information Loss at the Data Link Layer in Smart Networks
Poornima Madaraje Urs, A. Mallikarjuna Reddy, Mallikarjunaswamy Srikantaswamy, Umashankar Mynayakanahally Lakshminarayan · Engineering Technology & Applied Science Research · 2025
In smart networks, data are becoming increasingly complex, and enhancement methods are required to ensure data integrity and reliability. This paper proposes a novel IoT framework using machine learning for the prediction and mitigation of information loss at the data link layer, where conventional methods have many limitations. These methods cannot handle dynamic networking conditions and complex data traffic on any network, yielding smaller accuracy with a high false positive ratio. This work proposes a Machine Learning-based Information Loss Prediction Framework (ML-ILPF) using machine learning algorithms such as Support Vector Machine (SVM) and Long Short-Term Memory (LSTM) networks to overcome these challenges. These models analyze the historical data of the network to identify anomalies and predict possible loss. Compared to traditional methods, the proposed ML-ILPF outperformed both Static Threshold-Based Methods (STBM) and Basic Statistical Models (BSM) with an increase of 0.25% in accuracy and a reduction of 0.30% in false positives. This improvement shows real strength in the inclusion of machine learning in IoT frameworks toward smarter and more reliable network management. ML-ILPF is a promising solution that can help predict information loss at DLLS and improve the reliability and efficiency of smart networks, opening the window for more resilient IoT applications.