The rapid-fire proliferation of Internet of effects ( IoT) bias has introduced significant security challenges, particularly in large- scale and miscellaneous IoT networks. These systems are decreasingly susceptible to different cyber-attacks due to the decentralized nature and limited computational capabilities of individual IoT bumps. Traditional intrusion discovery systems( IDS) struggle to directly identify sophisticated and evolving attack patterns in similar surroundings. To address these limitations, this study proposes a new sequestration- conserving mongrel Convolutional intermittent Neural Network( CRNN) model integrated with allied literacy formulti-class intrusion discovery in IoT and Industrial IoT( IIoT) networks. Federated learning enables decentralized training of the model across multiple IoT bias without transferring raw data, thereby conserving data sequestration and icing compliance with data protection regulations. The cold-blooded CRNN armature leverages the strengths of Convolutional Neural Networks( CNNs) for point birth and intermittent Neural Networks( RNNs) for landing temporal dependences in network business. This combination significantly enhances the model’s capability to descry a wide range of attack types, including low- frequence and sophisticated pitfalls. The proposed model is trained and estimated using the Edge- IIoT dataset, demonstrating high performance with a discovery delicacy of 98.93. The results show balanced perfection and recall across all attack classes, including grueling orders similar as SQL Injection and Man- in- the- Middle attacks. This balance contributes to minimizing both false cons and false negatives, perfecting the overall trustability and robustness of the intrusion discovery system. By furnishing real- time discovery and sequestration- conserving training, the proposed approach offers a practical, scalable, and secure result acclimatized for complex IoT surroundings. It addresses critical gaps in being IDS fabrics by combining advanced deep literacy styles with allied literacy, paving the way for unborn secure and intelligent IoT deployments.