Design method of mu-miso hybrid precoding based on multi-agent deep reinforcement learning

Abstract

The invention discloses a MU-MISO hybrid precoding design method based on multi-agent deep reinforcement learning, which is suitable for downlink systems in communication. In this method, the base station constructs multiple deep reinforcement learning agents for calculating the simulated precoding matrix, each agent includes an action prediction network and an experience pool with priority, and each agent shares a centralized reward value prediction network and a centralized evaluation network to jointly explore analog precoding strategies. In this method, the base station obtains the channel state information of multiple users, inputs the user channel information into the constructed agent, and outputs the corresponding analog precoding matrix; and then calculates the digital precoding vector of each user through the zero-forcing precoding and water injection algorithm. Digital precoding matrix. It can effectively solve the problems of high complexity of hybrid precoding design and poor performance of attainable rate in massive MIMO system, and has strong robustness to channel environment.

Description

technical field [0001] The invention relates to a MU-MISO hybrid precoding design method, and is especially suitable for the MU-MISO hybrid precoding design method based on multi-agent deep reinforcement learning used in downlink systems in communication. Background technique [0002] Massive Multiple-Input Multiple-Output (MIMO), as an effective method to improve network transmission rate and energy efficiency, is regarded as one of the key technologies of the new generation wireless communication network. The MIMO system can make full use of space resources and double the system capacity without increasing spectrum resources and antenna transmission power. [0003] However, massive MIMO systems still face many challenges in practical applications. The design of beamforming matrices in mmWave systems is constrained by expensive RF hardware. The traditional all-digital beamforming structure needs to be equipped with a radio frequency link for each transmitting antenna and ...

AI Technical Summary

Problems solved by technology

[0005] Technical problem: Aiming at the deficiencies of the above technologies, provide a solution to the problem of high complexity of hybrid precoding design and poor attainable rate performance in massive MIMO systems, and has strong robustness to the channel environment. MU-MISO hybrid precoding design method for multi-agent deep reinforcement learning

Images