US20260238240
2026-08-13
Electricity
H04B1/0475
The patent application introduces a multi-head adaptive controller designed to enhance wireless communication systems by integrating Digital Pre-Distortion (DPD) and Self-Interference Cancellation (SIC) functionalities. This integration is achieved through a shared parameter neural network model, which outputs control signals to improve the efficiency and linearity of a transceiver's power amplifier. The system aims to support full-duplex communication, allowing simultaneous transmission and reception on the same frequency, which is crucial for modern wireless standards like 5G and 6G.
In wireless communications, DPD is a technique used to counteract the nonlinear distortions introduced by power amplifiers, while SIC is used to mitigate interference from a system's own transmissions. Traditionally, these functions are handled by separate actuators, each with its own computational resources and models. The application proposes a unified approach where a single actuator uses shared input signals and kernels to manage both DPD and SIC, significantly reducing the complexity and resources required.
The invention leverages a neural network with multiple heads, where each head is responsible for different tasks related to DPD and SIC. This shared architecture allows for a reduction in the number of parameters needed, leading to improvements in power consumption, memory usage, and overall system efficiency. By utilizing a common set of neurons and parameters, the system can effectively control both DPD and SIC operations, achieving better performance with fewer resources.
The integrated approach offers several benefits over conventional systems. It enables a significant reduction in the computational and physical resources required for DPD and SIC operations, leading to approximately a 50% decrease in both power consumption and memory footprint. This efficiency is particularly advantageous for mobile devices, where power and space are limited. Additionally, the shared parameter model facilitates easier implementation of full-duplex communication, enhancing the capabilities of next-generation wireless networks.
The application provides a detailed description of the system's architecture, including the use of a neural network with shared hidden layers and multiple output heads for DPD and SIC. This configuration allows the system to adaptively control the transceiver's operations, improving signal quality and reducing interference. The document emphasizes the flexibility of the approach, noting that it can be applied across various wireless standards and devices, from mobile phones to base stations, supporting both 3GPP and non-3GPP access technologies.