Design and implement an advanced deep learning model for real-time anomaly detection in industrial IoT sensor data. The solution should include data preprocessing, model architecture selection, training methodology, evaluation metrics, and deployment considerations. Address challenges including concept drift, limited labeled data, and edge deployment constraints.