基于隔离粒子群算法的智慧矿山5G SA网络切片部署策略

Slice deployment strategy for 5G SA network based on isolated PSO in intelligence mine

5G网络的发展促使采矿行业战略转型升级,降低采矿行业的安全风险,提高了采矿生产效率。智慧矿山5G SA(stand-alone)网络将用户面功能(UPF)下沉到矿区,控制面留在大区中心,远程控制、5G视频回传、VR/AR巡检和数据采集类等所有数据业务均通过专网访问矿区应用服务器。提出隔离粒子群算法(PSO_I),且引入贝叶斯评估方法计算粒子平均隔离因子,将其作为约束条件改善传统粒子群算法容易陷入局部最优的缺点。基于网络切片编排开销以及物理节点资源和链路资源的利用率构造隔离适值函数,通过迭代求解快速找到切片编排方案。经过矿区实际5G网络切片编排测试,结果显示PSO_I算法能够提升切片接受率、收益成本比和链路资源利用率,在降低执行时间的同时提高了网络上行速率,PSO_I算法优化的网络上行平均速率比SecPSO、Slice和SSPS算法分别提高19%、48%和52%。

The development of 5 G network has promoted the strategic transformation and upgrading of the mining industry,which can reduce the safety risks of the mining industry and improve the efficiency of mining production.The intelligence mine 5 G SA network sinks the user plane function to the mining area,and it leaves the control plane in the center of the large area.All data services such as remote control,5 G video backhaul,VR/AR inspection,and data collection are accessed through the 5 G edge computing interface to the mining area application servers.This paper proposes the PSO_I algorithm,and introduces the Bayes evaluation method to calculate the average particle isolation factor,which is used as a constraint to improve the shortcomings of traditional PSO.The isolation fitness function is constructed based on the network slicing orchestration overhead and the utilization of physical node resources and link resources,and the slicing orchestration scheme is quickly found by iterative solution.After the actual 5 G network slice arrangement test in the mining area,the results show that the PSO_I algorithm can improve the slice acceptance rate,the benefit-cost ratio and the link resource utilization rate,and improve the network uplink rate while reducing the execution time.Compared with the SecPSO,Slice and SSPS algorithms,the network uplink average rate optimized by the PSO_I algorithm is increased by 19%,48%and 52%.