Novel reinforcement learning method based on grid user position automatic antenna parameter adjustment

Abstract

The invention discloses a novel reinforcement learning method based on grid user position automatic antenna parameter adjustment. The method is used for a double-layer heterogeneous cellular network comprising a plurality of macro base stations, micro base stations and a plurality of macro users. In order to enable a plurality of users moving at a high speed in a complex network environment to always maintain a high weighted sum rate, the invention provides the novel reinforcement learning method based on grid user position automatic antenna parameter adjustment. The method comprises the following two steps: (1) in an offline modeling stage, achieving the biggest advantage that the time overhead and the calculation complexity during online learning can be reduced; and (2) in an online learning stage, based on the real-time SINR value fed back by a user, providing antenna parameter configuration capable of maximizing the weighted sum rate R of the user by using the provided novel reinforcement learning method. Compared with a traditional method, the method provided by the invention has the advantages that the applicable scene is closer to the current network condition and the reinforced learning has a good effect on time sequence prediction, and meanwhile, the method based on the grid user position has better expansibility.

Description

technical field [0001] The present invention relates to a method for automatic and joint adjustment of multiple antenna parameters. Specifically, it is a new type of deep enhancement for automatic antenna parameter adjustment based on gridded user location information in a double-layer heterogeneous cellular network with strong user mobility. A learning method belongs to the technical field of wireless communication. Background technique [0002] With the popularization of 5G, the existing network structure is gradually developing to a dense and complex topology. Followed by the continuous improvement of users' perception of network quality, traditional network coverage optimization methods have been difficult to meet the high-speed, low-latency network development process, and the adjustment of antenna parameters has a greater impact on network coverage . The current antenna parameter adjustment is mostly based on traditional manual parameter adjustment, but its shortcomi...

AI Technical Summary

Problems solved by technology

[0005] Aiming at the problems of poor expansibility of existing methods and insufficient consideration of complex existing network conditions, the present invention proposes a new reinforcement learning method based on gridded user position automatic antenna parameter adjustment

Images