This research proposes a new Federated Deep Learning Framework specifically designed to safeguard Internet of Things networks by effectively identifying Wormhole Attacks. The framework embarks on a comprehensive pre-processing journey that includes various strategies for handling missing data, implementing Winsorizing for outlier management, and utilising the Min-Max scaling technique for normalisation. As authors delve into the framework's core, an advanced feature extraction phase unfolds. This phase leverages a wide range of categories, such as Temporal Features extracted through a combination of Long Short-Term Memory networks and Feedforward Neural Networks, Statistical Features consisting of Median Absolute Deviation, mean, variance, skewness, and kurtosis, Database-based Features encompassing Packet Rate and Packet Size Distribution, Correlation-based Features obtained through Cross-Correlation analysis, and behavioral Features that reveal communication patterns, frequency of interactions, and deviations from everyday routines. The framework embarks on a comprehensive pre-processing journey that includes various strategies for handling missing data, implementing Winsorizing for outlier management, and utilising the Min-Max scaling technique for normalisation. As we delve into the framework's core, an advanced feature extraction phase unfolds. This phase leverages a wide range of categories, such as Temporal Features extracted through a combination of Long Short-Term Memory networks and Feedforward Neural Networks, Statistical Features consisting of Median Absolute Deviation, mean, variance, skewness, and kurtosis, Database-based Features encompassing Packet Rate and Packet Size Distribution, Correlation-based Features obtained through Cross-Correlation analysis, and Behavioral Features that reveal communication patterns, frequency of interactions, and deviations from everyday routines.