The presented slides

Download Report

Transcript The presented slides

Insert Title Here
Innovative Data Collection and
Dissemination in V2X Networks
Wassim Drira, Fethi Filali
4/13/2015
Connected
vehicles and road
safety
Context & Problem Statement
Multiple Data
Sources
GPS devices
Road Sensor (WaveTraf)
Crowd sourced data
Traffic
Analytics
MasarakTM Traffic
Information
Platform
Mobility
Analytics
Real-time traffic information
Routing & Navigation
Historical traffic data
Traffic incident information
Traffic forecast information
Origination-Destination Studies
Road work information
Parking/Public transit info
Weather information
Open APIs / SDKs for 3nd Parties
DATA COLLECTION + MANAGEMENT + ANALYTICS
Delivery:
Web
Mobile
SMS
Voice
FM
4
Intelligent Transportation Systems (ITS)
5
Data collection & dissemination in V2X
3G/LTE
IEEE802.11p
OBU
•
•
•
•
ADSL, Fiber, etc.
RSU
TMC
MSP
OBU: On-Board Unit
RSU : Road Side Unit
TMC : Traffic Management Center
MSP : Mobility Service Provider
Data Collection
Data Dissemination
6
Named Data Netorking (NDN)
•
NDN (Named Data networking) (http://named-data.net) is a future Internet
architecture based on:
– Treats content as a primitive – decoupling location from identity, security and access,
and retrieving content by name.
– Replace the where with what in the network.
– It can run alone or on top of IP stack.
– Hierarchical names
– Request/Response
– three main data structures:
• the FIB (Forwarding Information Base),
• Content Store (buffer memory)
• and PIT (Pending Interest Table)
7
Named Data Netorking (NDN)
8
NDN-Query: Motivations
•
Need to collect and analyze traffic information in order to:
–
–
–
•
•
Understand the traffic status
Detect incidents
Etc.
The obvious solution is to collect the periodic beacons
from the vehicular network through Road Side Units
(RSU) in a centralized Traffic Management Center (TMC).
But, this solution generate a tremendous amount of data
to be transmitted to TMC and then processed.
–
~ 114GB per day and per RSU
 The idea is to filter and aggregate the collected
data on the fly.
9
NDN-Query: Naming Pattern
query
OBU
RSU
TMC
MSP
response
+ Data is collected from one or many node.
+ Data is built on the fly and hop by hop.
+ Fully distributed query mechanism
+ Data aggregation hop by hop
10
NDN-Query: Step by Step example
11
Publish/Subscribe extension to NDN: Why?
•
•
•
•
•
•
•
•
Use of multicast and caching.
Decouple publishers and subscribers.
Push enabled dissemination.
Scalability: it is able to handle large number of publishers and subscribers and
large amount of information to disseminate.
Efficiency: content is not transmitted multiple times by a server or on a link.
Subscriber offline and mobility support.
Does not distinguish between transient or persistent data.
Tagging publications with version, generation time and freshness (new or
resubmitted).
12
Pub/Sub : Network Model
•
•
•
OBU : can act as subscriber, publisher or Local Rendezvous Node (L-RN).
RSU : can act as subscriber, publisher or L-RN.
TMC : hosts the Pub/Sub broker having local and Global Rendezvous Nodes (G-RN), powerful
servers, NDN routers and Back-up nodes.
o L-RN : Local-Rendezvous node receives events from nodes in layer one or two
o G-RN : Global-Rendezvous node receives events from nodes in layer two or three.
o BNs : Back-up Nodes are subscribed to all the content descriptors in order to store all the
publications and be able to provide this content to disconnected subscribers when needed.
OBU
•
•
•
•
RSU
TMC
OBU: On-Board Unit
RSU : Road Side Unit
TMC : Traffic Management Center
MSP : Mobility Service Provider
MSP
13
Pub/Sub : Content Identification
•
Content Identification
–
–
Content Name
Content Descriptor (CD)
•
•
•
•
/traffic/qatar/doha/accidents, /traffic/qatar/doha/jams, /traffic/qatar
Location
Time : Time Stamp, Version , Validity Time (Pub: finite or infinite VT,
Sub: infinite, finite, or associated to a unique publication)
Data Tables
–
–
–
Subscription Table (ST): stores subscriptions, number of subscribers, In/Out
interfaces, cache of recent publications and timers in RNs and on-route
nodes.
Publish-Forwarding-Table (PFT): On receiving a Publish message, only nonexisting messages in the PFT are forwarded and stored for a time T (the
round trip delay) in the same table.
Updates to CS: store publish messages and validity timers to allow on-route
caching and the automatic removing of expired messages.
14
Pub/Sub : Subscribe & Publish processing
15
Pub/Sub : What’s more?
• The extension allow
disconnected nodes to
get published messages
during this period when
they return.
• Pub/Sub supports
subscribers mobility
16
Simulation Setup
NDN-Q parameters
Publish/Subscribe parameters:
• VT=10s
• Publish VT=infinite
• T2=2s.
17
Simulation results : NDN-Q
18
Simulation results : Pub/Sub data dissemination
19
Simulation results : Pub/Sub data collection
20
Conclusions
•
We presented NDN-Q:
–
–
–
–
•
•
A fully distributed query mechanism that does not require knowledge of the network organization
and vehicles position.
A new query naming pattern and aggregation scripts.
The query response in NDN-Q is built on the fly and hop by hop.
The evaluation argues that this extension reduces the number of packets collected from data
sources, efficiently handles the mobility of vehicles, and deliver query results in a reasonable time
(280ms).
The Pub/Sub extension to NDN provides an Efficient Data Collection and Dissemination in
V2X Networks.
More information about our works in the NPRP project CoSMob can be foun in
www.cosmob.org
21