A network resource allocation method based on deep reinforcement learning

Abstract

The invention discloses a network resource allocation method based on deep reinforcement learning, which comprises the following steps: S1, establishing a CSCN-based downlink transmission link systemmodel, and calculating the data transmission rate of SBS by analyzing the data transmission rates of different links in the model used by a user; S2, proposing a game problem, and formulating the problem into a multi-agent non-cooperative game problem by using the game theory with the goal of maximizing the network throughput; S3, predicting the movement mode of the user by using an LSTM model, and selecting the user with the optimal transmission condition by the SBS according to the movement mode of the user and the cache state of the SBS connected with the user; S4, establishing an RL-LSTM framework to allow the SBS to complete the effective allocation of the channel resources. According to the method and the system, the movement mode of the user and the cache state of the SBS connectedwith the user are fully considered, so that the SBS selects the user with the optimal transmission condition, and the network throughput of the system is improved.

Description

technical field [0001] The present invention relates to a resource allocation method, in particular to a network resource allocation method based on deep reinforcement learning, and belongs to the field of channel resource allocation of novel cellular networks. Background technique [0002] With the widespread popularization and application of many mobile devices, in order to meet the demand for rapid growth of mobile data at this stage, relevant people in the industry have proposed a large number of new network architectures. [0003] Cache-based CSCN (Cloud-based small cell network, cloud small cell network CSCN) is currently one of the most promising architectures, which consists of a cloud pool and several SBS (small base stations, small base stations) connected to the cloud pool composition. Compared with the macro base station, the SBS has the advantages of low power, low price, and can flexibly supplement the blind area that the macro base station cannot cover. There...

AI Technical Summary

Problems solved by technology

[0004] The traditional random allocation method randomly allocates channels to users through SBS. Although this method can realize the allocation of network channel resources, the problem of low utilization of channel resources will occur in this way, which will lead to a decrease in user satisfaction.

Images