A method, device and medium for user clustering and power allocation based on deep q-network

Abstract

The invention discloses a user clustering and power allocation method, device and medium based on a deep Q network, using user clustering and power allocation to model a joint optimization problem; establishing a BP neural network to realize the power allocation function in the joint optimization problem ; Use the training data set to train the BP neural network, test the network and save the BP neural network model, obtain the power distribution results under different channel conditions, and realize the power distribution; model the user clustering problem as a reinforcement learning task; construct according to the reinforcement learning task Deep Q-network: After the network is trained online, the deep Q-network is trained according to the input state, and the best action is selected as the best clustering result to realize user clustering. The invention can reduce the online calculation complexity, guarantee user fairness to a certain extent and effectively improve the spectrum efficiency of the system.

Description

technical field [0001] The invention belongs to the technical field of resource allocation in a communication system, and in particular relates to a user clustering and power allocation method, device and medium based on a deep Q network. Background technique [0002] Facing the serious shortage of wireless spectrum resources and the current situation that the spectrum utilization rate of existing communication links is close to the limit, how to further improve the spectrum efficiency and system capacity to meet the requirements of large traffic, huge connections and high reliability under the full-scenario application of future wireless communication systems The demand is a key issue that research institutes in the field of wireless communication must urgently solve. Non-orthogonal and large dimensions are considered to be an effective way to improve the utilization of spectrum resources. In 2010, NTT DoCoMo of Japan proposed for the first time the power-domain Non-orthog...

AI Technical Summary

Problems solved by technology

However, the channel characteristics between different users are complex, and it is difficult for traditional methods to capture the potential relationship between users

At the same time, the solution space of the optimization problem is huge, and the nonlinear search process is inevitable

Therefore, it is difficult to obtain good user clustering and power allocation results using traditional methods, and the performance of the NOMA system is still greatly limited

The survey shows that the current research on large-scale MIMO-NOMA has not formed a systematic and comprehensive intelligent solution. The research focuses on a single perspective and the solidified deep learning network structure used has become the main reason for the limited system performance. Further technological breakthroughs are urgently needed

Images