AirComp Gradient Sign Alignment With Clipping and Bias
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing federated learning methods in wireless communication systems face challenges in aligning gradient signs without power loss or device loss during signal transmission, particularly in aligning gradient signs using AirComp in a range of signal amplitude of a receiver.
Innovation Solution
A method for aligning gradient signs in a wireless communication system involves obtaining a second flatten vector with gradient values, applying clipping and bias values, and transmitting it with channel information using AirComp, where the bias value is determined by the number of participating terminals and statistical characteristics, and phase compensation is applied.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If gradient values are transmitted directly using AirComp without clipping and bias application, then transmission process is simple, but gradient signs cannot be aligned and received signal distribution becomes wide causing power loss
Solution Approach 1:
The patent applies clipping and bias application to gradient values before transmission to align gradient signs in advance. This preliminary processing ensures that gradients with the same sign are clustered together, narrowing the received signal distribution and reducing power loss during AirComp transmission without requiring complex real-time adjustments
Solution Approach 2:
The patent transforms gradient values by changing their parameters through clipping (limiting value ranges) and bias application (shifting value distribution). This parameter transformation aligns gradient signs while maintaining the essential information needed for federated learning, achieving both signal concentration and information preservation
2Loss of energy
If clipping and bias values are applied to align gradient signs, then power efficiency is improved and signal distribution is narrowed, but processing complexity increases
Solution Approach 1:
The patent segments the gradient processing into distinct steps: clipping operation followed by bias application. This segmentation allows each operation to be performed independently and efficiently, reducing overall processing complexity while achieving gradient sign alignment and power efficiency
3Productivity
If all terminals participate in federated learning with direct transmission, then learning coverage is maximized, but signal amplitude variation causes reception power loss
Solution Approach 1:
The patent applies parameter changes (clipping and bias) to all participating terminals' gradient values, enabling them to transmit with aligned gradient signs. This allows all terminals to participate in federated learning while maintaining concentrated signal distribution and avoiding reception power loss, thus preserving both learning coverage and energy efficiency
Data Source
AI summary
Disclosed is a method for aligning gradient symbols by using bias regarding AirComp in a signal amplitude range of a receiver. A method for aligning gradient symbols according to an embodiment of the present disclosure comprises the steps of: performing iteration of gradient values included in a first planar vector and applying clipping and bias values so as to obtain a second planar vector having gradient values in which symbols are aligned; and transmitting the second planar vector together with channel information to a server in the form of AirComp. Various embodiments of the present disclosure may be linked to an artificial intelligence module, a drone (unmanned aerial vehicle, UAV), a robot, an augmented reality (AR) device, a virtual reality (VR) device, a device related to a 5G service, and the like.