A Distributed A-C Based Intelligent Access Control and Resource Allocation Method

Abstract

The invention relates to an intelligent access control and resource allocation method based on distributed A-C, which belongs to the technical field of communication. In this method, according to the performance requirements of eMBB and URLLC slices, a dual-objective optimization model that combines eMBB slice user transmission rate and URLLC slice user delay is constructed. Secondly, the federated learning and reinforcement learning are integrated to establish a federated reinforcement learning framework, and each agent cooperates to update the global weight parameters, and the local data of each agent is not exchanged with each other. The agents in the federated reinforcement learning framework use A-C learning to continuously interact with the environment, and each agent cooperates with each other, dynamically adjusts access control and resource allocation strategies, and optimizes the system model. The present invention can meet the performance requirements of users of each slice and maintain the user queues of each slice in a stable state, optimize the transmission rate of eMBB users and the time delay of URLLC users, and improve resource utilization.

Description

technical field [0001] The invention belongs to the technical field of communication, and relates to an intelligent access control and resource allocation method based on distributed A-C. Background technique [0002] The next-generation mobile communication network will support diversified business scenarios in a large number of vertical industries, such as smart security, telemedicine, smart home, autonomous driving, enhanced implementation, power inspection, etc. These application scenarios with different communication requirements usually include large-scale machine Massive Machine-Type Communication (mMTC), Enhanced Mobile Broadband (eMBB) and Ultra-Reliable and Low-Latency Communication (URLLC). Network slicing based on software-defined network (Soft Defined Network, SDN) and network function virtualization (Network Function Virtualization, NFV) technology refers to building the underlying physical network into multiple independent logical virtual networks. Network sli...

AI Technical Summary

Problems solved by technology

However, driven by the rapid development of wireless applications with diverse service requests, how to use the limited wireless network resources in the network to formulate radio access network (Radio Access Network, RAN) slices that provide specific services to meet different application requirements. Efficient and dynamic intelligent resource allocation strategies remain a challenging problem

[0003] At present, the existing technologies have the following problems: First, the existing resource allocation methods for network slicing are aimed at improving the revenue of infrastructure providers, without considering the characteristics of heterogeneous service requests in the network, and without specific analysis of the access slice users. Performance requirements, or for multiple network slicing scenarios, only a single performance is used to describe user requirements, but in fact, user performance requirements are different for different slices

Secondly, the existing access control and resource allocation problem regards the performance of users as a constraint and aims at maximizing revenue, ignoring the characteristics of dynamic arrival of services in the network, and not considering access control strategies to limit access to the network When the amount of random data arriving is too large, it may affect the stability of the network, and the static access control and resource allocation strategies are no longer applicable to the dynamically changing network environment

Images