Transcript V3: A Vehicle-to-Vehicle Live Video Streaming Architecture

V3:
A Vehicle-to-Vehicle Live Video
Streaming Architecture
Meng Guo; Ammar, M.H.;Zegura,E.W.;
Pervasive Computing and Communications, 2005. PerCom 2005. Third
IEEE International Conference on
08-12 March 2005 Page(s):171 - 180
Agenda
• Introduction
• Problem Description
• System Design
–
–
–
–
Triggering
Transmitting
Dynamics management
Dealing with multiple requests
• Simulations
• Conclusion
Introduction
V3 : An architecture to provide live video
streaming service to driving vehicles
Potential Applications :
1. Driver Assistance
2. Business Applications
3. Military Applications
Two Types of network :
1. Infrastructure-based
2. Vehicle-to-vehicle
Possibility
In terms of Bandwidth
IEEE 802.11
With high speed vehicles
54Mbps
1Mbps
In terms of Video bitrate
320 * 240, 65536 color
15fps, MPEG
100 – 150kbps
Challenges
1. Network partitions
•
•
Radio range
Penetration ratio
2. Dynamic conditions
3. Multiple mobile video sources
Problem Description
•
1.
2.
3.
4.
Based on mobility and the number of video
sources, we categorize these applications
into four classes:
Single stationary source (SSS)
Multiple stationary sources (MSS)
Single mobile source (SMS)
Multiple mobile sources (MMS)
Sample Scenario
1. A set of vehicles V drive on the road network N.
2. A receiver vehicle r ∈ V wishes to receive video
captured within destination region A
3. A sends a request Q = (r, t0, l0,A,D) towards the destination region.
Architecture Design Overview
Video Trigger Messages
Video Transmission
Triggering
Directed Flooding
Reason:
1.
The overhead is tolerable
2.
Shorter delay and highly reliability
3.
A flooding based approach enables continuous triggering,
while unicast based approach does not.
Algorithm
1.
A vehicle first calculates if it is in the trigger message
forwarding zone (TMFZone).
2.
If so, forwards the trigger message
3.
Else, buffers the trigger message, and re-calculates its
location every α time units.
TMFZone
d(A,v) / d(A, r) : distance from vehicle v and r to the destination
After boundary condition : Tc – Ts > Td – Tc + o
Current time
Trigger message start time
Deadline
time
Video Source Selection
• If there are multiple vehicles in the destination
region simultaneously, only one of them
needs to be triggered, based on largest tin.
Current location
Exit point
Continuous Trigger
• Pre-trigger :before the request message
reaches the destination region, vehicles that
receive the request become “potential video
sources”
• Post-trigger : a vehicle keeps forwarding the
trigger message even after it already leaves
the destination region.
Video Transmission
• Ensures vehicles are in Data Forwarding Zone
(DFZone)
• Multiple forwarders approach
Source location
Current location
Current time
Video
data
generated
Dynamics
• Receiving vehicles drive away
• New vehicles joined in
Simulation
Traffic Software Integrated
System (TSIS)
Each Direction has 4 /5
lanes
Total length : 9miles
6 entrances / exits
Traffic Situation
Dense : average vehicle-vehicle distance is much
smaller than a vehicle’s radio range, and the vehicles
can still maintain normal speed
Medium : average vehicle-vehicle distance is similar to
the vehicle’s radio range
Sparse : vehicle-vehicle distance is larger than the
vehicle’s radio range
Congested : vehicles drive significantly slower (10mph
to 15mph) than normal speed.
Results (1)
Location-based – TMFZone Approach
Direction-based – vehicles that are driving towards the
destination region forward the video trigger message
Destination-based - only vehicles that will drive through
the destination region forward the trigger message.
Results (2)
Viewing Delay - delay between the receiver vehicle’s video playback
time and the video generation time.
Service Delay - delay between the time when a receiver vehicle sends
out the trigger message and the time when the receiver vehicle first
receives the video data.
Results (3)
Conclusion
• Propose and design V3
• Define Zone for transmission
• Pre-trigger and post-trigger method
presented
• Simulations verify flooding approach