Bayesian Network Prediction of Mobile User Throughput in 5G Wireless Networks
Qingmin Meng, Xiaoqiang Fang, Wenjing Yue, Yang Meng, Jingcheng Wei
School of Communication and Information Engineering
Nanjing University of Posts and Telecommunications, Nanjing 210003, China
e-mail: mengqm@njupt.edu.cn, yuewjg@njupt.edu.cn, 15195880838@163.com, 18896585748@163.com,
wjcyx2009@qq.com
Abstract—By combining artificial intelligence and machine
learning, next-generation cellular systems will enable advanced
data analysis techniques to achieve efficient service quality
management and network automation. In this paper, the
Bayesian Network (BN) is used for the reliability prediction of
throughput. The forecast is to predict future test results
through parameter estimation. In the studied Bayesian
network learning stage, the load of the base station, user
location and moving speed all affect the user’s received signal-
to noise ratio (SNR) and signal interference plus noise ratio
(SINR). The test result is the probability or number of times
that throughput of a low-speed mobile user satisfies the
threshold. Computer simulation results show that the model
can well infer the user’s throughput under low speed
movement conditions.
Keywords-Bayesian network; inference; learning; cellular
systems; throughput
I. INTRODUCTION
The fifth generation (5G) wireless network will
support a variety of new service types, such as strict
reliability services and strict delay services [1]. The
cellular system with the deployment of various kinds of
base stations is one of wireless network structures where a
static topology and centralized management are usually
adopted. The design of the cellular system need to face
many system design issues, such as high-capacity
resource allocation under various kinds of interference.
Since frequency reuse is a common resource allocation
method in cellular design, it is possible for different cells
to use the same channels, which results in inter-cell co-
channel interference (CCI). Compared to previous cellular
systems, the effect of CCI to 5G systems using dense
networks and small-cells is even greater. Next, we discuss
regional coverage related to both CCI and mobility. This
is because the interference power received by mobile
devices is different in different locations. User
movements may be roughly classified as three movement
types, i.e., low, medium, and high velocities. User
movement may result in reduced Quality of Service(QoS)
performance. Therefore, we need to consider how
interference and mobility affect the 5G QoS performance.
The solutions to these problems will help the design of
service strategies to improve network reliability. Multiple
service policy operations can effectively support specific
services, such as service secure quality indication (SSQI)
under certain load setting at edge of 5G network and
ensuring sufficient bandwidth reservation for priority
resource users. Compared with existing 4G cellular
networks, running a large number of QoS service
strategies requires advanced 5G network techniques with
enhanced intelligence [1]. The difference from the
existing received signal strength
indication (RSSI) is that SSQI will jointly consider the
impact of CCI, mobility and load.
In recent years, the research of artificial intelligence
(AI) and machine learning has received extensive
attention in many fields, but there is not much research
work on network automation for 5G-like cellular
networks. The work in [2] discusses the progress in self-
organization and network automation of 5G wireless
networks. For wireless access networks, the work in [3]
studies the application of machine learning and artificial
intelligence in short-term resource management strategies.
Based on deep reinforcement learning, the work in [4]
proposes an intelligent strategy optimization method in
wireless network management. Moreover, the distributed
resource sharing from device to device in a heterogeneous
network is characterized as a Bayesian coalition formation
game [5]. The Bayesian Reinforcement Learning (RL)
model can be used to improve the performance of
resource allocation schemes. For 5G access networks, the
work in [6] proposes a network reliability service model
and Bayesian network based causal inference. For
handover between different networks, the work in [7]
proposes a wireless network availability prediction based
on dynamic Bayesian networks (DBN). Related work in
[8] studied a smart cooperative spectrum sensing
algorithm based on a nonparametric Bayesian learning
model. As telecom companies are adopting new network
technologies like visualization, Software Defined Network-
Network Function Virtualization (SDN-NFV) and Mobile
Cloud Computing, AI and machine learning is going to
play a big role in smooth integration of these technologies
and automating the networks. The recent work in [9]
proposes a machine learning framework for resource
allocation and discusses how to integrate supervised
learning into cloud computing assisted cellular systems to
291
The 10th International Conference on Communications, Circuits and Systems
978-1-5386-9388-9/18/$31.00 ©2018 IEEE
下载后可阅读完整内容，剩余4页未读，立即下载
weixin_38667403
- 粉丝: 2
- 资源: 915
我的内容管理
展开
最新资源
- IPQ4019 QSDK开源代码资源包发布
- 高频组电赛必备：掌握数字频率合成模块要点
- ThinkPHP开发的仿微博系统功能解析
- 掌握Objective-C并发编程：NSOperation与NSOperationQueue精讲
- Navicat160 Premium 安装教程与说明
- SpringBoot+Vue开发的休闲娱乐票务代理平台
- 数据库课程设计：实现与优化方法探讨
- 电赛高频模块攻略：掌握移相网络的关键技术
- PHP简易简历系统教程与源码分享
- Java聊天室程序设计：实现用户互动与服务器监控
- Bootstrap后台管理页面模板（纯前端实现）
- 校园订餐系统项目源码解析：深入Spring框架核心原理
- 探索Spring核心原理的JavaWeb校园管理系统源码
- ios苹果APP从开发到上架的完整流程指南
- 深入理解Spring核心原理与源码解析
- 掌握Python函数与模块使用技巧
资源上传下载、课程学习等过程中有任何疑问或建议，欢迎提出宝贵意见哦~我们会及时处理！
点击此处反馈
