Frequency spectrum resource management and distribution method based on federated learning

Abstract

The invention discloses a frequency spectrum resource management and allocation method based on federated learning. The method comprises the steps: enabling the terminal equipment to collect the state information from an environment; if the information change exceeds a threshold value, making a decision of a frequency spectrum, a sub-channel and transmission power of channel transmission; enabling the terminal equipment to collect the state information and reward information at the next moment after the decision, and adopting a gradient descent method to train an actor network and a reviewer network in the main network; when the accumulated gradient of the terminal equipment reaches a preset gradient threshold value, uploading the main network parameters of the terminal equipment to the base station; after the quantity of the main network parameters collected by the base station reaches a threshold value, aggregating the main network parameters according to a preset weight and broadcasting to all the terminal equipment; and enabling the terminal equipment to update the parameters of the main network and the target network after receiving the aggregation parameters. In order to solve the problems of low spectrum utilization rate, easy leakage of user privacy and the like in the prior art, the invention adopts an asynchronous federated learning mode, and has the advantage of low communication consumption while solving the problems.

Description

Technical field [0001] The present invention relates to a spectral resource management allocation method, in particular to a spectral resource management allocation method based on federal learning. Background technique [0002] In future communication networks, the growing growth and diversified service types of terminal equipment are constantly emerging, and network capacity is huge. Even with full spectrum, there is still a challenge in the face of huge bandwidth requirements. Only the utilization of the full range can only be effectively alleviated, and the network load can be effectively eased to maximize user service quality. However, relative to the existing wireless communication system, the full spectrum network spectrum resource presents differential and high dynamics, and the existing spectrum resource management mechanism based on mathematical optimization theory lacks forecasting capabilities for available spectrum resources, so it is difficult to effectively apply. ...

AI Technical Summary

Problems solved by technology

Due to the wide distribution of data in the real environment, the base station collects these scattered data for centralized learning to complete resource scheduling. The danger of deriving private information such as the location of the secondary user and the communication frequency band from the sent information

Images