5G slice resource scheduling method

Abstract

The invention provides a 5G slice resource scheduling method. The method comprises the following steps: initializing a memory pool and learning parameters of a deep reinforcement network; the memory pool comprises a network state, an action, a reward value and a next network state; the action represents a 5G slice resource allocation strategy; feeding collected historical distribution data of 5G slice resources into a deep neural network for reinforcement learning, and forming a tetrad by a network state, an action, a reward value and a next network state obtained by each time of learning; randomly selecting a plurality of tetrads and feeding the tetrads into the neural network for training to obtain a slice resource allocation model; in response to a received terminal service request, performing 5G slice resource allocation based on the slice resource allocation model; according to the method, rapid model training can be carried out by adopting limited samples, so that the complexity of slice resource allocation model calculation of massive users in various communication scenes is greatly reduced, the decision duration of slice resource allocation is effectively shortened, and the slice resource allocation efficiency is improved.

Description

technical field [0001] The present invention relates to the technical field of mobile communication, in particular to a method for scheduling 5G slice resources. Background technique [0002] As one of the key technologies of 5G (5th Generation Mobile Communication Technology, the fifth generation mobile communication technology), network slicing can create and maintain multiple customized independent logical networks on the basis of shared physical network facilities, targeting different types of 5G The form of customized dedicated network slicing for specific application scenarios improves the heterogeneity, flexibility, scalability, profitability, and security of future network services. [0003] At present, most 5G slice scheduling methods use discrete particle swarm optimization or genetic algorithm to achieve algorithm optimization. However, in the face of a large number of user requests, using the above algorithm to calculate slice resource allocation has the problem...

AI Technical Summary

Problems solved by technology

However, in the face of a large number of user requests, using the above algorithm to calcula

Images