【OAI】OAI5G核心网VPP-UPF网元分析
文章目录
- VPP_UPF_CONFIG_GENERATION.md
- VPP UPF Configuration Generation
- Environment variables
- Interfaces
- Interface Configuration Examples
- Central UPF
- A-UPF
- I-UPF
- UL CL
- FEATURE_SET.md
- VPP_UPG_CLI
- 参考文献
VPP_UPF_CONFIG_GENERATION.md
VPP UPF Configuration Generation
这个文章比较关键
VPP_UPF_CONFIG_GENERATION.md
init.conf
startup_debug.conf
upf_profile.json
create_configuration.py
The VPP-UPF configuration for different UPF deployment options (A-UPF, I-UPF, ULCL) are not compatible with each other.
The example described in VPP_UPF_CONFIG shows a central UPF node with one N3 and one N6
interface.
To ease the deployment, the UPF configuration can be auto-generated using the create_configuration.py
script. It is
located in scripts/upf_conf/create_configuration.py
. If you choose to use the official OAI docker images (from v1.5.0), this script is
executed upon start of the containers.
It takes three positional arguments:
usage: create_configuration.py[-h][--rename]init_pathprofile_pathstartup_path
The init_path
is the path to the init file that will be generated. A json file is written to the profile_path
which
contains the UPF profile which is sent to the NRF.
The startup_path
requires an existing VPP startup_debug.conf
file with placeholders (@value@
). These will be
overwritten with the corresponding environment variables.
If you choose to rename using the --rename
option, the host interfaces are renamed from e.g. eth1
to n6-3
.
Which interface is renamed depends on the configured subnet.
Environment variables
The UPF configuration is built entirely from environment variables. There are static variables, which always have to be
present:
Environment Variable | Description | Used in |
---|---|---|
NAME | Used to build the UPF FQDN | init.conf, profile.json |
MNC | Used to build the UPF FQDN | init.conf, profile.json |
MCC | Used to build the UPF FQDN | init.conf, profile.json |
REALM | Used to build the UPF FQDN | init.conf, profile.json |
VPP_MAIN_CORE | Which CPU core ID to use for the main thread | startup_debug.conf |
VPP_CORE_WORKER | Which CPU core ID to use for the worker thread * | startup_debug.conf |
SNSSAI_SD | SD from the SNSSAI | profile.json |
SNSSAI_SST | SST from the SNSSAI | profile.json |
DNN | DNN | profile.json |
VPP_PLUGIN_PATH | Path where the UPG VPP plugin is installed | startup_debug.conf |
* This config from VPP allows to configure more than one worker thread, e.g. 1-4, but it is not recommended in the
VPP-UPF, as there are race conditions.
The UPF FQDN is built as follows:
<name>.node.5gcn.mnc<mnc>.mcc<mcc>.<realm>
The NAME variable is translated to lower-case.
Then, there are support features which you also have to configure:
Environment Variable | Description |
---|---|
REGISTER_NRF | If the UPF should register to NRF |
NRF_IP_ADDR | The NRF IP address |
NRF_PORT | The NRF Port |
HTTP_VERSION | HTTP version towards NRF (1/2) |
Apart from this global configuration, you can dynamically configure your interfaces.
The syntax to configure an interface is as follows:
IF_<Index>_<Value>
Each interface can have multiple values, whereas the index describes the interface.
The following values are available:
Value | Description | Mandatory |
---|---|---|
IP | IP address of the interface** | yes |
TYPE | Either N3, N4, N6 or N9 | yes |
NWI | Network Instance of the interface | yes (not for N4) |
DNAI | Data Network Access Identifier | no |
IP_REMOTE | Next hop IP, optional when using --rename | yes/no |
** Should belong to subnet of existing host interface
An example of a UPF configuration is as follows (see also the docker-compose file):
- IF_1_IP=192.168.70.201
- IF_1_TYPE=N4
- IF_2_IP=192.168.72.201
- IF_2_TYPE=N3
- IF_2_NWI=access.oai.org
- IF_3_IP=192.168.73.201
- IF_3_TYPE=N6
- IF_3_IP_REMOTE=192.168.73.135 # EXT-DN IP Address
- IF_3_NWI=internet.oai.org
- NAME=VPP-UPF
- MNC=95
- MCC=208
- REALM=3gppnetwork.org
- VPP_MAIN_CORE=0
- VPP_CORE_WORKER=1
- VPP_PLUGIN_PATH=/usr/lib/x86_64-linux-gnu/vpp_plugins/ # Ubntu18.04
- SNSSAI_SD=123
- SNSSAI_SST=222
- DNN=default
- REGISTER_NRF=yes
- NRF_IP_ADDR=192.168.70.130
- NRF_PORT=80
- HTTP_VERSION=1
Interfaces
We use veth pairs to configure the UPF interfaces. Each UPF interface (e.g. N3) has a host interface IP and the IP address
used for signaling. You need one subnet for each UPF interface. In a typical deployment, you need at least three subnets:
- Signaling (N4)
- Access (N3)
- Core/Internet (N6).
The IP
value should belong to the same subnet as the host interface, but must not be the same.
The IP_REMOTE
value also needs to be in the same subnet and is used to setup the routes accordingly.
When you use the --rename
option, the IP_REMOTE
is automatically calculated from the subnet from the IP
value.
The remote IP is always the first host in the subnet, e.g. for the IP 192.168.70.201
in the subnet 192.168.70.0/24
,
the remote IP is 192.168.70.1
.
The value you provide overrides the default value. When you do not rename the interfaces, you need to manually specify
the remote IP address.
Interface Configuration Examples
The dynamic creation allows you to configure different UPF types.
Central UPF
This UPF has one N3, one N4 and one N6 interface:
- IF_1_IP=192.168.70.201
- IF_1_TYPE=N4
- IF_2_IP=192.168.72.201
- IF_2_TYPE=N3
- IF_2_NWI=access.oai.org
- IF_3_IP=192.168.73.201
- IF_3_TYPE=N6
- IF_3_NWI=internet.oai.org
A-UPF
This UPF has one N9, one N4 and one N6 interface:
- IF_1_IP=192.168.70.202
- IF_1_TYPE=N4
- IF_2_IP=192.168.74.202
- IF_2_TYPE=N9
- IF_2_NWI=iupf.node.5gcn.mnc98.mcc208.3gppnetwork.org
- IF_2_DNAI=iupf
- IF_3_IP=192.168.75.202
- IF_3_TYPE=N6
- IF_3_NWI=internet.oai.org
- IF_3_DNAI=internet
When you want to use multiple UPFs which should be chained or form a graph, you need to ensure that the NWI
of the
configuration matches the FQDN
of the other UPF. In this case, IUPF
should have the following FQDN config:
- NAME=IUPF
- MNC=98
- MCC=208
- REALM=3gppnetwork.org
I-UPF
This UPF has one N3, one N4 and one N9 interface:
- IF_1_IP=192.168.70.203
- IF_1_TYPE=N4
- IF_2_IP=192.168.72.203
- IF_2_TYPE=N3
- IF_2_NWI=access.oai.org
- IF_2_DNAI=access
- IF_3_IP=192.168.74.203
- IF_3_TYPE=N9
- IF_3_NWI=aupf.node.5gcn.mnc98.mcc208.3gppnetwork.org
- IF_3_DNAI=aupf
As you can see, the N9 interface is in the same subnet as the N9 interface of the I-UPF example. This means
that the SMF will create an edge between these two nodes (as long as the FQDN/NWI configuration matches).
Also, the DNAI is configured for these examples to enable SMF in selecting a specific UPF for a PDU session.
UL CL
This UPF has one N3, one N4 and two N9 interfaces:
- IF_1_IP=192.168.70.204
- IF_1_TYPE=N4
- IF_2_IP=192.168.72.204
- IF_2_TYPE=N3
- IF_2_NWI=access.oai.org
- IF_2_DNAI=access
- IF_3_IP=192.168.74.204
- IF_3_TYPE=N9
- IF_3_NWI=aupf1.node.5gcn.mnc98.mcc208.3gppnetwork.org
- IF_3_DNAI=aupf1
- IF_4_IP=192.168.76.204
- IF_3_TYPE=N9
- IF_3_NWI=aupf2.node.5gcn.mnc98.mcc208.3gppnetwork.org
- IF_4_DNAI=aupf2
FEATURE_SET.md
FEATURE_SET.md
VPP_UPG_CLI
VPP-UPF 可以进行的一些操作
$ docker exec -it vpp-upf ./bin/vppctl show upf helpadf show upf adf commandsapplication show upf application <name>applications show upf applications [verbose]association show upf associationbihash show upf bihash <v4-tunnel-by-key | v6-tunnel-by-key | qer-by-id | peer-index-by-ip> [detail|verbose]flows show upf flowsflow show upf flow commandsgtpu show upf gtpu commandsnode-id show upf node-idnwi show upf nwipfcp show upf pfcp commandspolicy show upf policyproxy show upf proxysession show upf sessionspecification show upf specification commandstdf show upf tdf commands
- Show PFCP session at UPF
$ docker exec -it vpp-upf ./bin/vppctl show upf session
CP F-SEID: 0x0000000000000001 (1) @ 192.168.70.133
UP F-SEID: 0x0000000000000001 (1) @ 192.168.70.202PFCP Association: 0TEID assignment per choose ID
PDR: 1 @ 0x7f6cef01cd00Precedence: 0PDI:Fields: 0000000dSource Interface: AccessNetwork Instance: access.oai.orgLocal F-TEID: 218138074 (0x0d0085da)IPv4: 192.168.72.202UE IP address (source):IPv4 address: 12.1.1.2SDF Filter [1]:permit out ip from any to assigned Outer Header Removal: GTP-U/UDP/IPv4FAR Id: 1URR Ids: [1] @ 0x7f6ceefde850QER Ids: [] @ 0x0
PDR: 2 @ 0x7f6cef01cd80Precedence: 0PDI:Fields: 0000000cSource Interface: CoreNetwork Instance: core.oai.orgUE IP address (destination):IPv4 address: 12.1.1.2SDF Filter [1]:permit out ip from any to assigned Outer Header Removal: noFAR Id: 2URR Ids: [1] @ 0x7f6ceefde8e0QER Ids: [] @ 0x0
FAR: 1Apply Action: 00000002 == [FORWARD]Forward:Network Instance: core.oai.orgDestination Interface: 1
FAR: 2Apply Action: 00000002 == [FORWARD]Forward:Network Instance: access.oai.orgDestination Interface: 0Outer Header Creation: [GTP-U/UDP/IPv4],TEID:9acb0442,IP:192.168.72.141
URR: 1Measurement Method: 0002 == [VOLUME]Reporting Triggers: 0001 == [PERIODIC REPORTING]Status: 0 == []Start Time: 2022/09/02 07:29:30:476vTime of First Usage: 0.0000 vTime of Last Usage: 0.0000 VolumeUp: Measured: 0, Theshold: 0, Pkts: 0Consumed: 0, Quota: 0Down: Measured: 0, Theshold: 0, Pkts: 0Consumed: 0, Quota: 0Total: Measured: 0, Theshold: 0, Pkts: 0Consumed: 0, Quota: 0Measurement Period: 10 secs @ 2022/09/02 07:29:40:476, in 5.597 secs, handle 0x00000c01
参考文献
参考文献1:OAI 基础核心网部署:oai-cn5g-fed/docs/DEPLOY_SA5G_BASIC_DEPLOYMENT.md
参考文献2:oai-cn5g-upf-vpp
参考文献3:UPD_DEPLOY_HELP
参考文献4:vpp-upf-docs
相关文章:

【OAI】OAI5G核心网VPP-UPF网元分析
文章目录 VPP_UPF_CONFIG_GENERATION.mdVPP UPF Configuration GenerationEnvironment variablesInterfacesInterface Configuration ExamplesCentral UPFA-UPFI-UPFUL CL FEATURE_SET.mdVPP_UPG_CLI参考文献 VPP_UPF_CONFIG_GENERATION.md VPP UPF Configuration Generation …...

【上进小菜猪】使用Ambari提高Hadoop集群管理和开发效率:提高大数据应用部署和管理效率的利器
📬📬我是上进小菜猪,沈工大软件工程专业,爱好敲代码,持续输出干货,欢迎关注。 介绍 Hadoop是一种开源的分布式处理框架,用于在一组低成本硬件的集群上存储和处理大规模数据集。Ambari是一种基…...

Day3--C高级3
一.编写一个名为myfirstshell.sh的脚本,它包括以下内容。 1、包含一段注释,列出您的姓名、脚本的名称和编写这个脚本的目的 2、和当前用户说“hello 用户名” 3、显示您的机器名 hostname 4、显示上一级目录中的所有文件的列表 5、显示变量PATH和HO…...
第9章 CURD操作与MemoryCache缓存的强制清理的实现
1 重构 Data.Repository<TEntity> using Core.Caching; using Core.Domain; using Core.Events; using Microsoft.EntityFrameworkCore; namespace Data { ///<typeparam name"TEntity">泛型类型实例(这里特指:1个指定实体的类型实例)。</typepa…...

TCP 协议特性详解
TCP 协议特性总结 TCP协议特点TCP协议段格式TCP原理确认应答(安全机制)超时重传(安全机制)连接管理(安全机制)(面试高频题)三次握手四次挥手 滑动窗口(效率机制)流量控制(…...
电子招投标采购系统源码:采购过程更规范,更透明
满足采购业务全程数字化, 实现供应商管理、采购需求、全网寻源、全网比价、电子招 投标、合同订单执行的全过程管理。 电子招标采购,是指在网上寻源和采购产品和服务的过程。对于企业和企业主来说,这是个既省钱又能提高供应链效率的有效方法…...

一篇了解智慧网关
智慧网关是指基于互联网技术的智能网关,能够连接不同的物联网设备和传感器,实现数据采集、信息传递、远程控制、通信管理等功能。作为物联网架构中的核心设备之一,智慧网关在智能家居、智慧城市、智能制造、智能交通、智能农业等领域得到了广…...

自学软件测试,从10K到40K的技术路线,也就是这些东西...
如果有一天我从梦中醒来时,发现自己的几年自动化测试工程师经验被抹掉,重新回到了一个小白的状态。我想要重新自学自动化测试,然后找到一份自己满意的测试工作,我想大概只需要6个月的时间就够了,如果比较顺利的话&…...

Qt libqrencode二维码——QtWidgets
前言 之前写过二维码的程序,但是在U盘上,没带,又重新找的网上资料写的。 网上二维码的生成,大多用到是第三方库libqrencode,这也一样: 效果图 本来是个动图的,都被和谐了,所以换成截图&…...

KDZD绝缘子表面电导盐密度测试仪
一、简介 智能电导盐密测试仪,也称为直读式等值盐密度测试仪,专为测试智能电导盐密度而设计。系统内置智能电导盐密度计算公式,读数直观。 人机交互采用真彩TFT液晶屏,操作简单,测试参数和结果一目了然。仪器自带微型打…...
如何降低电动汽车软件的开发成本和风险?
大多数的汽车制造商无法从销售电动汽车(EV)中获得利润,但计划快速进入市场的电动汽车初创公司是无法承担这样的损失的。 由于飙升的电池价格、高昂的组件成本和低迷的销量削弱了盈利能力,电动汽车初创公司必须将视线转到软件开发…...

使用pytest和allure框架实现自动化测试报告优化
目录 -x出现一条测试用例失败就退出测试 生成测试报告json pytest: 需要安装pytest和pytest-html(生成html测试报告) pip install pytest 和 pip install pytest-html 命名规则 Pytest单元测试中的类名和方法名必须是以test开头,执行中只能找到test开头…...

chatGPT免费站点分享
下面的应该都能用,试试吧... ChatGPT是一种人工智能聊天机器人,能够生成虚拟语言和交互回复。使用ChatGPT,您可以与机器人进行真实的交互,让机器人根据您提出的问题或请求来生成回复。但是,在使用ChatGPT时࿰…...
【计算机网络】已知一个/27网络中有一个地址是 167.199.170.82,问这个网络的网络掩码,网络前缀长度和网络后缀长度是多少?网络前缀是多少?
目录 题: 1. 网络掩码 2.网络前缀长度 3.网络前缀 4.网络后缀长度 题: 已知一个/27网络中有一个地址是 167.199.170.82, 问这个网络的网络掩码,网络前缀长度和网络后缀长度是多少?网络前缀是多少? 1.…...
Java8 - Stream
介绍 Java8中有两大最为重要的改变。第一个是 Lambda 表达式; 另外一个则是 Stream API(java.util.stream.*)。Stream是 Java8中处理集合的关键抽象概念,它可以指定你希望对集合进行的操作,可以执行非常复杂的查找、过滤和映射数据等操作。使用 Stream …...

什么样的冷链保温箱,既环保又实用?
冷链物流运输已经应用在了很多行业中,作为冷链物流运输中的重要设备——冷链保温箱,起到了举足轻重的作用。如果选择不当,选到了劣质产品,尤其是化学行业或者食品行业,就有可能造成试剂失效或者是影响粮食食品安全问题…...

Eclipse的介绍与安装
Eclipse简介 Eclipse 是一个开放源代码的,基于 Java 的可扩展开发平台。Eclipse官方版是一个集成开发环境(IDE),可以通过安装不同的插件实现对其它计算机语言编辑开发,如C、Php、Python等等。 Eclipse的下载 下载时需要访问网址 http://…...

<IBM AIX> 《AIX中HA的网卡IP确认方法》
《HA网卡确认方法》 1 HA IP地址类型2HA IP地址设置原则3 HA网卡日常查看方法3.1 查看hosts文件3.2 查看网卡和IP 4 通过命令方式直接查看5 直接查看HA配置4 HA网卡SMIT查看方法(暂不完整)4.1 根据hosts文件确认IP对应的别名4.2 根据serviceIP确认Persis…...

AMB300系列母线槽红外测温解决方案某锂电厂房项目案例分享
安科瑞 耿敏花 一、 行业背景 近年来,在国家政策引导与技术革新驱动的双重作用下,锂电产业保持快速增长态势,产业规模持续扩大,同时新能源产业工厂锂电池生产线对于电的依赖性很高,因而对供电设备的可靠性提出…...
go语言学习——4
文章目录 Map Map map是一种特殊的数据结构:一种元素对pair的无序集合,pair的一个元素是key,对应一个value,这种结构也称为“关联数组或字典”引用类型 map声明 var map1 map[keytype] valuetype var map1 map[string] intmap可以…...

MFC内存泄露
1、泄露代码示例 void X::SetApplicationBtn() {CMFCRibbonApplicationButton* pBtn GetApplicationButton();// 获取 Ribbon Bar 指针// 创建自定义按钮CCustomRibbonAppButton* pCustomButton new CCustomRibbonAppButton();pCustomButton->SetImage(IDB_BITMAP_Jdp26)…...

8k长序列建模,蛋白质语言模型Prot42仅利用目标蛋白序列即可生成高亲和力结合剂
蛋白质结合剂(如抗体、抑制肽)在疾病诊断、成像分析及靶向药物递送等关键场景中发挥着不可替代的作用。传统上,高特异性蛋白质结合剂的开发高度依赖噬菌体展示、定向进化等实验技术,但这类方法普遍面临资源消耗巨大、研发周期冗长…...

SCAU期末笔记 - 数据分析与数据挖掘题库解析
这门怎么题库答案不全啊日 来简单学一下子来 一、选择题(可多选) 将原始数据进行集成、变换、维度规约、数值规约是在以下哪个步骤的任务?(C) A. 频繁模式挖掘 B.分类和预测 C.数据预处理 D.数据流挖掘 A. 频繁模式挖掘:专注于发现数据中…...

转转集团旗下首家二手多品类循环仓店“超级转转”开业
6月9日,国内领先的循环经济企业转转集团旗下首家二手多品类循环仓店“超级转转”正式开业。 转转集团创始人兼CEO黄炜、转转循环时尚发起人朱珠、转转集团COO兼红布林CEO胡伟琨、王府井集团副总裁祝捷等出席了开业剪彩仪式。 据「TMT星球」了解,“超级…...

【Oracle】分区表
个人主页:Guiat 归属专栏:Oracle 文章目录 1. 分区表基础概述1.1 分区表的概念与优势1.2 分区类型概览1.3 分区表的工作原理 2. 范围分区 (RANGE Partitioning)2.1 基础范围分区2.1.1 按日期范围分区2.1.2 按数值范围分区 2.2 间隔分区 (INTERVAL Partit…...
rnn判断string中第一次出现a的下标
# coding:utf8 import torch import torch.nn as nn import numpy as np import random import json""" 基于pytorch的网络编写 实现一个RNN网络完成多分类任务 判断字符 a 第一次出现在字符串中的位置 """class TorchModel(nn.Module):def __in…...

uniapp 开发ios, xcode 提交app store connect 和 testflight内测
uniapp 中配置 配置manifest 文档:manifest.json 应用配置 | uni-app官网 hbuilderx中本地打包 下载IOS最新SDK 开发环境 | uni小程序SDK hbulderx 版本号:4.66 对应的sdk版本 4.66 两者必须一致 本地打包的资源导入到SDK 导入资源 | uni小程序SDK …...
人工智能--安全大模型训练计划:基于Fine-tuning + LLM Agent
安全大模型训练计划:基于Fine-tuning LLM Agent 1. 构建高质量安全数据集 目标:为安全大模型创建高质量、去偏、符合伦理的训练数据集,涵盖安全相关任务(如有害内容检测、隐私保护、道德推理等)。 1.1 数据收集 描…...
全面解析数据库:从基础概念到前沿应用
在数字化时代,数据已成为企业和社会发展的核心资产,而数据库作为存储、管理和处理数据的关键工具,在各个领域发挥着举足轻重的作用。从电商平台的商品信息管理,到社交网络的用户数据存储,再到金融行业的交易记录处理&a…...

【深度学习新浪潮】什么是credit assignment problem?
Credit Assignment Problem(信用分配问题) 是机器学习,尤其是强化学习(RL)中的核心挑战之一,指的是如何将最终的奖励或惩罚准确地分配给导致该结果的各个中间动作或决策。在序列决策任务中,智能体执行一系列动作后获得一个最终奖励,但每个动作对最终结果的贡献程度往往…...