Tone Reservation Configuration for Machine-Learned PAPR Reduction

The patent applies preliminary action by pre-training machine learning models offline to generate PAPR reduction signals. The models are trained beforehand on extensive datasets and then deployed for real-time operation, where they can quickly generate reduction signals without requiring complex real-time computation. This shifts the computational burden from runtime to training time, resolving the contradiction between effectiveness and computational complexity.

The patent uses copying by creating simplified versions of complex PAPR reduction algorithms through machine learning models. Instead of implementing complex heuristic algorithms during transmission, the system learns optimal reduction patterns during training and copies these patterns into compact model structures that can be efficiently deployed, maintaining effectiveness while reducing runtime computational complexity.

The patent applies segmentation by dividing the reserved resources into specific patterns with predetermined structures. Instead of arbitrarily allocating resources, the system defines segmented patterns (e.g., contiguous subcarriers, periodic patterns) that can be efficiently indicated through compact signaling. This structured segmentation reduces the number of bits needed to describe resource allocations while maintaining effective PAPR reduction.

The patent uses parameter changes by representing reserved resource allocations through a small set of key parameters (e.g., starting position, length, pattern type) rather than detailed bitmaps. By changing the representation from exhaustive resource indication to compact parameter-based description, the system achieves effective resource reservation with minimal signaling overhead.

The patent applies taking out by extracting the PAPR reduction function from the data-carrying subcarriers and placing it in dedicated reserved resources. When reserved resources overlap with physical channels, the system extracts the necessary information from overlapping channels and processes them separately, ensuring that PAPR reduction signals are generated without causing interference to channel data, thus maintaining resource utilization while avoiding signal distortion.

The patent uses an intermediary approach by introducing a separate processing stage for handling overlaps between reserved resources and physical channels. The system identifies overlapping regions and applies specific handling mechanisms (such as rate matching or puncturing) as intermediary steps between resource allocation and signal transmission, preventing direct interference while maintaining efficient resource utilization.