Page History

Setup the Physical Machine:

• Ubuntu 20 (ubuntu-20.04.1-desktop-amd64)

1. Setup an hypervisor
   
   1. verify virtualization enabled
   2. install kvm/libvirtd
2. Create a network for VMs
3. Create 4 VMs (at start 3 VMs are enough)
   
   1. 192.168.122.11 - Master
   2. 192.168.122.101 - Vehicles/RSU(s)/BS(s) 
   3. 192.168.122.111 - Edge-1
   4. 192.168.122.211 - Edge-2
4. Test VMs connectivity

Setup a Master and two worker nodes (links: link-1, link-2, etc)

1. disable swap (persistently)
2. setup the Multus as a CNI (using this link-1)
3. note down the available IP-ranges for container applications:
   1. range: w.x.y.z/a

Explore the networking plugin options that can be used in our proposed scenarios of vehicles:

• Multus (multiple virtual interfaces)
• OpenShift-SDN CNI
• Links
  • Multus link
  • Openshift-SDN link
• etc

DSRC range:

• RSU - 300m (link)
• Vehicle - 15m (link)

Routing considerations with DSRC/LTE both:

• SDN-based routing application
• Multi-hop approach (using tables)

1. Architecture - figures/explanation
2. Components - figures/explanation
3. Scenarios - figures/explanation
   1. simple diagram to understand the scenarios
   2. technical aspect of the proposed scenarios
4. V2X Communication - figure/explanation
5. Implementation - plan/approach
   1. container vehicles
   2. maps (Jeju/Udo islands)
   3. proximity services
   4. broadcast content
   5. epc on the edge sites
   6. location prediction model
   7. sdn
6. Relevant Blueprints
7. Contributors
8. References
9. Ask Questions - table of questions
10. Tasks

Vehicle-info container

(start/boot-priority = 1)

1. Design the database
2. Develop the scripts to create the database
3. Containerize it into the docker hub
4. Setup a (dbeaver) plugin to access container database
   
   1. Windows
   2. Ubuntu
5. Expose the port (or make it accessible on some portal)

Map container

(start/boot-priority = 2)

1. Downloaded .osm/.mbtile formatted files for Jeju/Udo Island
2. GIS-based maps
   1. set locally
   2. containerize it (Folium-based map server)
   3. find .shp for Jeju/any city/area
   4. use wgs84/epsg4326 format (geojson)
   5. traffic data for vehicles (Jeju/any)
   6. pandas, geopandas, folium
   7. graphical analysis of traffic (map/graphs)
   8. interactive layers (i.e. ___)
3. Setup the OpenStreetMap (PostgreSQL). Steps to setup locally.
   
   1. on local system
   2. containerize it
   3. make it accessible to the cars
4. Analyze/correct the .osm maps for Jeju/Udo Islands
5. Search ways for creating roads, intersection etc
6. Implement roads. intersection etc
7. Explore implementation options to dynamically move container vehicles
   1. Consider the RSUs/BSs in this architecture

Vehicle container(s)

(start/boot-priority = 3)

Specifications:

1. Vehicle Class
2. Attributes
   
   1. name
   2. curr_location [long., lat.]
   3. junction_location [long., lat.]
   4. curr_velocity [m/s]
   5. direction [N/E/S/W/NE/ES/SW/WN]
   6. possible_moves [N/E/S/W/ES/SW/WN]
   7. net_d2d
   8. net_d2b
   9. net_d2p
   10. range_of_net_d2d
   11. etc.
3. Methods:

   • turn

     Inputs	Outputs	Actions/Effects	Description
     vehicle id direction	next location	decrease velocity decelerate by some factor while turning update the location#

     
     

   • accelerate

     Inputs	Outputs	Actions/Effects	Description
     vehicle id direction acceleration factor	new location	increase velocity update the location#

     
     

   • broadcast

     Inputs	Outputs	Description
     vehicle id bits	none	The udp client will be responsible to broadcast the traffic to the gateway. The proximity/other service at the edge node is responsible to decide on which edge nodes does the traffic need to be broadcasted

     
     

   • park/stop

     Inputs	Outputs	Actions/Effects	Description
     		decrease velocity to zero update the location#

     
     

   • update the location

     Inputs	Outputs	Actions/Effects	Description

     
     

   • sync. location to proximity server

     Inputs	Outputs	Actions/Effects	Description
     vehicle id location	response of success/failure	vehicle location must be updated on the proximity server	updates the current location to server after every 2 seconds.

     
     

   • etc.

     
     

Method details:

1. Turn (to describe)
2. Accelerate (to describe)
3. Broadcast (to describe)
4. Park/Stop (to describe)
5. Update the location
6. Sync. the location to proximity server
7. etc.

1. Junctions
2. Locations
3. Road track points
4. etc.

1. Specify the criteria of updating locations of container vehicles
2. Describe the the selected standard approach for implementing proximity service

1. OpenShift
2. µONOS
3. others

1. Explore the OpenShift-SDN CNI
2. Routing logic as a Northbound application
3. Apply routing to: 
   1. vehicle container network interfaces
   2. RSU container network interfaces
4. Test the validity of routing

1. East-west communication
2. Mechanism to share information

1. Finding datasets
   
   1. real-time
   2. simulated link-1, link-2, link-3
2. Processing/Analyzing datasets
3. Model details
   1. architecture?
   2. models:
      1. model A:
         1. input(s):
         2. output(s):
      2. model B:
         1. input(s):
         2. output(s):
      3. model C:
         1. input(s):
         2. output(s):
      4. model D:
         1. input(s):
         2. output(s):
      5. model E:
         1. input(s):
         2. output(s):
4. Training
   1. training/testing data split
   2. training details
5. Testing
   1. testing time
   2. testing analysis
6. Validation/Verification
   1. define the validation metrics
   2. define the verification mechanism
7. Exposing model REST API(s)
   1. define the endpoints
      1. define/document the JSON formats required
8. Integration (basic)
9. Integration (testing functionality)

1. Docker-compose
2. Helm charts

1. Installation of basic packages:
   1. script to install kvm/libvirt
   2. script to install ansible
   3. script to install vagrant
2. Provision the infrastructure using:
   1. vagrant (primary option)
   2. ansible (secondary option)
3. To setup K8s cluster - using Ansible scripts:
   1. write a script to download/start 3 VMs
   2. write a script to install master/worker nodes node remotely
   3. write a script to setup the CNI plugin (if possible, configured with the chosen SDN controller)
4. To bring up the applications - using Helm-charts/Docker-compose:
   1. ml model
   2. sdn controller
   3. container mapserver
   4. container proximity- service
   5. container vehicle
   6. routing ml application
5. To deploy the SDN Northbound application - using ...