Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

Page Tree
expandCollapseAlltrue
rootMEC-based Stable Topology Prediction for Vehicular Networks

MEC-based Stable Topology Prediction for Vehicular Networks

Use Case Details:

Attributes

Description

Informational

TypeNew
Blueprint FamilyIntegrated Cloud Native (ICN)ICN BP family intends to address deployment of workloads in a large number of edges and also in public clouds using K8S as resource orchestrator in each site and ONAP-K8S as service level orchestrator (across sites).  ICN also intends to integrate infrastructure orchestration which is needed to bring up a site using bare-metal servers.
The use case is proposed under the ICN BP family
Industry sector

Area: SDN & NFV

Company: ATTO Research

University:

Jeju National

Gachon University
Country: Republic of Korea

ATTO RESEARCH focuses
We focus on the problems related to networking and software technology for a better connection.
It has created a
The technology that allows to build IT infrastructure, and aims to grow into a ‘Software Defined Infrastructure’ company.
Business
driversIn order
driverThe stable vehicular network is essential to enable various applications such as autonomous driving through VANETs
, stable vehicular network is essential
. Proposed MEC architecture
proves
tends to
be
enable a promising infrastructure where a stable network topology can be predicted locally
in order
to improve the network performance by providing intensive calculation for vehicles in the adjacent roads. Thus, converging the two concepts of MEC
(using ICN)
and topology prediction can provide a strong use
-
case for the vehicular networks such as proactive path stabilization.The MEC-based Efficient Routing Algorithm
is able to
can provide a stable path by using the predicted future position for the nearby vehicles. The information can also be made available to the adjacent road resulting
to be
from being useful to provide a stable topology on the road tracks.
Use cases
  • Stable Network Topology Prediction in IoV at the Edge
  • Proactive D2D Connectivity for Inter/Intra Cell Communication
The topology prediction performed at the edge will provide real-time/proactive network topology and a stable routing path in IoV.Business cost (initial build cost target objective)
Business use case
  • Edge cloud deployable at RSUs to support applications such as ML-based location prediction, topology stabilization

Business cost - Initial build

Minimal

The minimal

configuration is three servers in total:

  • Master/Database node (
1st
  • 1st server)
  • Edge node 1 (
2nd
  • 2nd server)
  • Edge node 2 (3rd server)
Scale & Type3 (x86 server)The physical machines are used for different purposes such as Master, Edge and Database nodes. 1st server will be used as a Master and Database node. The remaining two physical servers will be used as two different edge nodes for computing.Nodes (virtual)
  • 1 master node
  • 1 database node
  • 2 edge nodes
Our system has three basic nodes. Master node being the one to manage the other nodes helps out in scaling up the number of edge nodes and applications on them. Edge nodes have the prediction models which provide the network topology in IoV. Database node contains the information of each edge node accessible to the Master node via which Master node can also view the overall network topology in IoV for different purposes.ApplicationsApplications for autonomous vehicles such as topology and location prediction on edge clouds.These applications (being latency sensitive) can be extended to any type of business use-cases.Infrastructure orchestration
  • ONAP4K8s (Service Orchestrator)
  • K8s Cluster
  • Ubuntu 18.04.3 LTS (OS)
For orchestration purpose, we will use ICN Blueprint's provided infrastructure. We will setup the edge locations on our local physical servers and provide with the performance test results.Workload typeContainers/Virtual-machinesAdditional details
  • Introduced intelligence at the Edge
  • Reducing backhaul due to Edge
  • Support of path within Single operator domain
PPT has been attached as a proposal statement.Summary/ Overview

Due to recent advancement in networks, IoV has a major role to play and is critical to performance. To this end, we introduce a machine learning (ML) approach integrated at the edge as a proposal under the ICN Blueprint Family.

As it is a known fact that the vehicles at the edge are directly connected to the Radio Station Units/Base Stations (RSU/BS) and placing such application/software packages at the edge can significantly enhance the performance of multi-hop device to device communication in the IoV domain.

Further, the extended version of Kalman Filter is proposed to predict the network topological locations. The inference engine used for predicting the vehicular locations and further forming them into predicted topology provides a stable path for the vehicles to communicate on the provided channels. Overall, the solution and architecture are flexible in nature which further allow the support of multiple use cases in 5G.

We also make the use of standard services such as Proximity Services, LTE Service, and our proposed Stable Path algorithm integrated along with the others.

  • Abbas, Muhammad Tahir, Afaq Muhammad, and Wang-Cheol Song. "Road-aware estimation model for path duration in internet of vehicles (IoV)." Wireless Personal Communications 109.2 (2019): 715-738.
  • Abbas, Muhammad Tahir, et al. "An adaptive approach to vehicle trajectory prediction using multimodel Kalman filter." Transactions on Emerging Telecommunications Technologies (2019): e3734.
  • Abbas, Muhammad Tahir, Afaq Muhammad, and Wang-Cheol Song. "SD-IoV: SDN enabled routing for internet of vehicles in road-aware approach." Journal of Ambient Intelligence and Humanized Computing (2019): 1-16.
  • Huang, Cheng, et al. "Virtual mesh networking for achieving multi-hop D2D communications in 5G networks." Ad Hoc Networks 94 (2019): 101936.

Motivation:

Motivational aspects are:

  • The topology in a vehicular network is updated and retrieved frequently
    • This causes path instability
  • Vehicular networks are wireless in nature
    • However, Software-defined networking (SDN) is originally designed for wired networks
  • Leads to the need for topology stability in vehicular networks

To this end, we introduce:

  • Computation at the Edge 
  • Topology prediction to proactively stabilize the paths in vehicular network

Motivational IdeaImage Removed

QoS-based multi-channel Scenario:

The architectural view of our system is as follows:

Image Removed

Stable Path Scenarios:

The below figure (shown left side) is a scenario where the devices are under same cell and in coverage, which want to communicate. The below figure (shown right side) is a scenario where the devices are under same cell and one oft them is out of coverage, which want to communicate. 

Image RemovedImage Removed

The below figure explores the Device to Device communication where a device 1 wants to communicate device-4 and both of them reside under a different Cell node. In this case, the packet is forwarded from one edge vEPC to another. This way the connectivity is provided as follows:

Image Removed

The below scenario explains the same as immediate previous scenario but the only difference among them is that one of the device is out of coverage and proximity service plays the role of provisioning the location of such devices.

Image Removed

The below figure shows the direct communication between the devices which reside under different Cells. The below path was provided as a result of proactive approach that we propose in our system. The major difference between the below two scenarios is that in such cases, we don't require the involvement of Edge vEPC in order to forward the packet to another road segment, thus reducing the latency and improving the performance in terms of bandwidth and efficient resource usage of the cellular spectrum. Both the devices are in coverage.

Image Removed

The below figure also shows the direct communication between the devices which reside under different Cells. The below path was provided as a result of proactive approach that we propose in our system. The difference from previously mentioned scenario is that one of these devices are out of coverage. In this case also, the benefits to reduce the latency and efficient resource usage were achieved.

Image Removed

Contributors:

Price factor depends on the cost of RSU quality, and should be only considered for physical deployment. i.e. wireless or wired.
Business cost - Operational

Virtual environment does not require cost.


Operational need

Using the frontend GUI to:

  • Orchestrate virtual resources
Manage the edge applications

Additional details
  • Support of path within a Single operator domain

PPT is attached as proposal statement.

Species Details:

Attributes

Description

Informational

Type

Integrated Cloud Native NFV/App stack (ICN)


Blueprint Family

Existing


Use case

Stable network topology in IoV


Blueprint proposed nameMEC-based Stable Topology Prediction for Vehicular Networks

Initial POD cost

Satellite POD


Scale & Type

System will be developed/deployed in VMs.


Applications
  • ML model
  • LTE network services
  • ProSe Functions

Open Air Interface (OAI) provided LTE network services will be used.

Infrastructure orchestration

  • OpenStack latest/stable release – VM orchestration
  • Kubernetes-based container orchestration
  • WeaveNet -based Container Networking
  • VNF Orchestration – ONAP
  • OS – Ubuntu 18.X LTS
  • CICD - Jenkins 2.249.1 LTS

SDN

ONOS will be used at the application layer


Workload type

VMs and Containers


Additional

  • 4 Virtual Machines
    1. One Orchestration node
    2. Three Edge nodes
  • Jenkins for Continuous Integration and Continuous Delivery
  • Personal servers will be integrated with the Linux Foundation servers

Committers and PTL (Project Technical Lead)

Please enter in all names of the committers for the project.

PTL is done off of self nomination process.  If you wish to be considered for the PTL, please indicate that by putting a Y in the self nomination column (use the slide to move the table left to right).  Per Akraino rules, if there is only one nominee, that person becomes PTL (when confirmed by the Akraino TSC).  If there is more than nominee, we will then have an election.
The election process is open and will go through 7 Oct. 2020 at Noon Pacific time. 

Committer

Committer Company

 Committer Contact InfoTime Zone

Committer Bio

Committer Picture

Self Nominate for PTL (Y/N)

Gachon University

malikasifmahmoodawan@gmail.com

Asia/Seoul (UTC+9)

Image Added

Y

First (1st)Second (2nd)

from:

to:

from:

to: Present


Gachon University


Asia/Seoul (UTC+9)

from:  

to: Present























CompanyEmail (Contact)ProfilesPictureAsif MehmoodJeju National Universitymalikasifmahmoodawan@gmail.comAfaq MuhammadJeju National Universityafaq@jejunu.ac.krSaqib MuhammadJeju National Universitysaqib@jejunu.ac.krMuhammad Ali Jibran---alijibran35@gmail.comWang-Cheol SongJeju National Universityphilo@jejunu.ac.krTaekyung LeeATTO Researchtaekyung.lee@atto-research.com


Info
titleHelp Us Improve the Wiki

This Wiki is owned by the Akraino Community. Contributions are always welcomed to help make it better!

In upper right, select Log In. You will need a Linux Foundation Account (can be created at https://identity.linuxfoundation.org/) to log-in. For a Wiki tutorial, please see Confluence OverviewThank you!



Recent space activity

Recently Updated
typespage, comment, blogpost
max5
hideHeadingtrue
themesocial