Adaptive Receiver Interference Classification Using Machine Learning

The patent applies parameter changes by transitioning from classical fixed-parameter interference rejection methods to machine learning-based adaptive parameter adjustment. The receiver uses trained machine learning models that dynamically adjust processing parameters based on the specific interference conditions detected, enabling optimal performance under varying co-channel and self-interference scenarios while maintaining manageable complexity through pre-trained models.

The patent substitutes classical mechanical/deterministic interference rejection algorithms with machine learning-based cognitive systems. Instead of fixed mathematical operations, the receiver employs trained neural networks or other ML models that learn optimal interference cancellation strategies from training data, replacing traditional signal processing mechanics with adaptive intelligence that handles complex interference patterns more effectively.

The patent applies preliminary action by pre-training machine learning models offline before deployment in the receiver. The complex training process that would otherwise burden the receiver is performed in advance using extensive training data and computational resources, allowing the receiver to only execute the trained model during operation. This shifts the complexity burden from the receiver to the training phase, enabling performance improvement without proportionally increasing receiver complexity.

The patent introduces an intermediary training system that bridges the gap between complex ML algorithms and the receiver implementation. The training system processes raw training data and generates optimized model parameters, acting as an intermediary that prepares the complex intelligence externally so the receiver can benefit from improved performance without bearing the full complexity burden of model training and adaptation.

The patent applies universality by designing a single machine learning-based receiver architecture that handles multiple interference types through one unified system. The receiver uses a general-purpose ML model that can be trained on diverse interference scenarios (co-channel interference, self-interference, and other conditions), allowing one device to adaptively handle various interference types without requiring separate specialized processors for each interference condition.

The patent applies dynamics by implementing an adaptive receiver that dynamically adjusts its interference rejection strategy based on real-time conditions. The machine learning model continuously processes incoming signals and adapts its parameters based on detected interference characteristics, enabling the system to transition between handling different interference types as conditions change, rather than being fixed to a single rejection method.