PhD Defense ExamPhD Defense Exam
Data Aggregation and Dissemination in Vehicular Ad-Hoc Networks
Khaled Ibrahim
Advisor: Dr. Michele C. Weigle
Computer Science Department
Old Dominion University,
Norfolk, VA 23529
February 21, 2011 1
OutlineOutline
• Introduction
• Motivation
• Problem Definition
• CASCADE– Local View Component– Extended View Component– Data Security Component– Data Dissemination Component
• Summary 2
IntroductionIntroduction
• What is a Vehicular Ad-Hoc Network (VANET)?
3
IntroductionIntroduction
• Communication Models In VANET:
• Vehicle-to-Vehicle (V2V)
• Vehicle-to-Infrastructure (V2I)
• Hybrid of V2V and V2I
4
IntroductionIntroduction
• Assumptions
• Transceiver
• GPS (D-GPS)
• Set of Public/Private Key Pairs
• Tamper-Proof Device
• Laser Rangefinder
5
MotivationMotivation
• VANET Applications:
• Safety Applications
• Informational Applications
• Entertainment Applications
Collision Warning
Congestion Notification
Music/Movie Sharing
6
MotivationMotivation
• Data Needed by VANET Applications:
• Common Data• Vehicle Location• Vehicle Speed
• Application Specific Data• Collision Location• Congestion Location• Songs/Movies to be shared
Collision Warning
Congestion Notification
Music/Movie Sharing
7
MotivationMotivation
• The Common Data Characteristics:
• Refresh or update rate
• Accuracy
• Volume
• Each category of applications needs a customized version
8
MotivationMotivation
• The Scalability Problem Example• N1 Safety Applications• N2 Informational Applications• N3 Entertainment Applications
• N1*10 + N2*3 + N3* 1
• 10 + 3 + 1 (Better Solution)
• 10 (The Best Solution)9
Problem DefinitionProblem Definition
How to securely and efficiently provide each VANET application with a customized version of the vehicular data based on its category.
10
CASCADECASCADE
CASCADECluster-based Accurate Syntactic Compression of Aggregated Data in VANETs
11
CASCADECASCADE
• Major Framework Components
• Local View
• Extended View
• Data Security
• Data Dissemination
12
CASCADECASCADE
Local View
Receiving Aggregated FrameBroadcasting Aggregated FrameReceiving Primary FrameBroadcasting Primary FrameData Flow in CASCADE
13
ContributionsContributions
• a lossless data compression technique based on differential encoding that has compression ratio of 86%• a syntactic data aggregation mechanism that can represent the vehicular data in a local view of length 1.5km in one single MAC frame
Local View Component
Data Dissemination ComponentExtended View Component
Data Security Component
14
• a probabilistic data dissemination technique that alleviates the spatial broadcast storm problem and effectively uses the bandwidth to disseminate data to distant areas in a short amount of time in addition to having less redundancy and reasonable coverage than other techniques.• a mechanism for recovering from the communication discontinuity problem in short time based on the traffic density in the opposite direction
• an investigation of the possible data structures for representing the vehicular data in a searchable format• a parametric mechanism for matching the vehicular data and providing a customized version of the data that satisfies certain characteristics based on the parameter value
• a light-weight position verification technique that quickly detects false data with very low false positives
CASCADECASCADE
15
Local View ComponentLocal View Component
16
Local View ComponentLocal View Component
• What is Local View?
• Local View Component Responsibility?
• Maintain an accurate Local View
• Add new vehicle
• Update vehicles locations
• Delete out of scope vehicles
17
Local View ComponentLocal View Component
• Local View Component Responsibility?
• Compress and aggregate the vehicular data in the local view and compose one aggregated frame that fits into a single MAC frame (2312 B)
18
Local View ComponentLocal View Component
19
• Data Compression– Differential coding– CASCADE-Max
• Vehicular Data Compression X (5 Bits) Y ( 7 Bits) Speed (5 Bits)
• Compression ratio is 86%
Local View ComponentLocal View Component
• What is the cluster dimension?
• Smallest aggregated frame
• Longest local view length
20
Local View ComponentLocal View Component
• Determined best cluster size experimentally• Cluster sizes
– Cluster length (62m,126m, 254m and 510m)– Cluster width (1 lane, 2 lanes, 4 lanes )
• Vehicular densities– low, medium and high
• Vehicular distribution– worst distribution (uniform distribution)– best distribution (clustered distribution)– expected distribution
21
Local View ComponentLocal View Component
22
Local View ComponentLocal View Component
• Local View Component:
• Maintain an accurate view for the traffic ahead for short distances (1.5 km)
• Compress and aggregate the local view data to fit into a single MAC frame
23
Extended View ComponentExtended View Component
24
Extended View ComponentExtended View Component
• Extended View Component Responsibility?
• Build and maintain the extended view
• Customize the extended view based on the predefined settings for each registered application.
25
Extended View ComponentExtended View Component
26
Extended View ComponentExtended View Component
• Build and Maintain Extended View
• Determine if two vehicles match
• Determine if two intersecting regions match
27
Extended View ComponentExtended View Component
• Determine if two vehicles match
• What threshold of difference for two vehicles should we accept as matching?
• Evaluated experimentally through simulation
• To maximize true positive and true negative and minimize false positive and false negative, use vehicle difference threshold of 16%
28
Extended View ComponentExtended View Component
• Determine if two intersecting regions match
• Does the data structure used to represent the regions matter?
• implemented comparison with graph structure and KD Tree structure
• KD Tree is 22% faster than graph, but uses 39% more memory
29
Extended View ComponentExtended View Component
• Customize the extended view
• Matching percentage - % of vehicles in the intersecting regions that match
• What matching % is required to accept the received aggregated frame?
30
Extended View ComponentExtended View Component
• What matching % is required to accept the received aggregated frame?
• Small matching %
• more aggregated frames will be accepted
• longer extended view
• may be less accurate
31
Extended View ComponentExtended View Component
• What matching % is required to accept the received aggregated frame?
• Large matching %
• fewer aggregated frames will be accepted
• shorter extended view
• may be more accurate
32
Extended View ComponentExtended View Component
33
Matching percentage threshold vs. extended view length
Safety Applications
Informational Applications
Entertainment Applications
Extended View ComponentExtended View Component
• Extended View Component:
34
• Build and maintain an extended view with maximum accuracy
• Customize the extended view based on the application settings (refresh rate, accuracy, view length)
Data Dissemination Data Dissemination ComponentComponent
35
Data Dissemination Data Dissemination ComponentComponent
• Disseminate data to distant areas in a short amount of time in addition to having less redundancy and reasonable coverage than other techniques
• Recover from the communication discontinuity problem in short time based on the traffic density in the opposite direction
36
Data DisseminationData Dissemination
• Broadcast
• DSRC 300 m
A
37
Data DisseminationData Dissemination
• Re-broadcast
– Flooding [Ni –MOBICOM’99]
– Weighted p-Persistence [Wisitpongphan-IWC’07]
– Slotted 1-Persistence [Wisitpongphan-IWC’07]
– Slotted p-Persistence [Wisitpongphan-IWC’07]
– Inter-Vehicle Geocast (IVG) [Bachir –VTC’03]
38
Data DisseminationData Dissemination
• Re-broadcast
– Inter-Vehicle Geocast (IVG)
•
• i is the message sender
• j is the message receiver
• Dij is the distance between vehicle i and vehicle j
• Tij is the re-broadcast timer
R
DRTT ij
ij
*max
39
Data DisseminationData Dissemination
• Re-broadcast
– Probabilistic- IVG (p-IVG) •
•
R
DRTT ij
ij
*max
densityp
1
40
Data DisseminationData Dissemination
• p-IVG Evaluation– Metrics
• MAC Delay• Reception Rate• Backoff Percentage• Dissemination Delay and Hop Count• Redundancy Factor• Coverage Percentage
41
Data DisseminationData Dissemination
Because using p-IVG reduces the media contention, the reception rate increases
42
Data DisseminationData Disseminationp-IVG takes less time to send the messages further using smaller number of hops
43
Data DisseminationData Dissemination
• Redundancy Factor– The optimal case is to receive each message once
redundancy factor = 0– Realistically1 the minimum redundancy factor = 0.4
[1] S.-Y. Ni, Y.-C. Tseng, Y.-S. Chen, and J.-P. Sheu, “The broadcast storm problem in a mobile ad hoc network,” in Proceedings of ACM Mobicom, Seattle, WA, Aug. 1999, pp. 151–162.
44
Data DisseminationData Dissemination
• Coverage %– Definition: % of vehicles within the transmission
range that received the message or any of its rebroadcast.
– The optimal dissemination technique should have 100% coverage.
45
Data DisseminationData Dissemination
IVG
46Number of Extra Copies
Data DisseminationData Dissemination
p-IVG
47Number of Extra Copies
Data DisseminationData Dissemination
• p-IVG Summary
• It can disseminate data to distant areas in a short amount of time in addition to, having less redundancy and reasonable coverage than IVG.
48
Data Dissemination Data Dissemination ComponentComponent
• Communication Discontinuity
• We have been assuming that the distance between any two communicating vehicles will not be greater that 250m.
• Removing this assumption results in possible breaks in communication
49
Data Dissemination Data Dissemination ComponentComponent
Sparse Traffic Clustered Traffic 50
Data Dissemination Data Dissemination ComponentComponent
• Yah rab
51
Ext
ende
d V
iew
Len
gth
(km
)
Data Dissemination Data Dissemination ComponentComponent
On-Demand Vehicular Gap-Bridging (OD-V-GB)
Broadcasting GBR Messages
Handling Received Aggregated Frames
On Demand Broadcasting
52
Data Dissemination Data Dissemination ComponentComponent
On-Demand Vehicular Gap-Bridging (OD-V-GB)
Handling Received Aggregated Frames
• Background process to build an extended view for the opposite direction (2 sec aggregated frames repository)
• Matching Percentage Threshold is 0%
53
Data Dissemination Data Dissemination ComponentComponent
On-Demand Vehicular Gap-Bridging (OD-V-GB)
Broadcasting GBR Messages
• Timer to track the most recent message received from traffic ahead
• If timer expires Discontinuity or Gap detected
• Then send a GBR request
54
Data Dissemination Data Dissemination ComponentComponent
On-Demand Vehicular Gap-Bridging (OD-V-GB)
On Demand Broadcasting
• Once they get in contact with a vehicle in the direction requesting help, they broadcast their opposite direction extended view in one aggregated frame
• What is the impact of the vehicular density? 55
Data Dissemination Data Dissemination ComponentComponent
56
Ext
ende
d V
iew
Len
gth
(km
)
Data DisseminationData Dissemination
• OD-V-GB Summary:
• It can recover from the communication discontinuity problem in short time based on the traffic density in the opposite direction
57
Data Dissemination Data Dissemination ComponentComponent
58
SummarySummary
• Local View Component
• a lossless data compression technique with compression ratio of 86%
• a syntactic data aggregation mechanism that can represent the vehicular data in a 1.5km area in single MAC frame
59
SummarySummary
• Extended View Component
• an investigation of the possible data structures for representing the vehicular data in a searchable format
• a parametric mechanism for matching the vehicular data and providing a customized extended view
60
SummarySummary
• Data Security Component
• a light-weight position verification technique that quickly detects false data with very low false positives
61
SummarySummary
• Data Dissemination Component
• a probabilistic data dissemination technique that • alleviates the spatial broadcast storm problem
• disseminate data to distant areas in a short amount of time in addition to having less redundancy and reasonable coverage than other techniques.
• a mechanism for recovering from the communication discontinuity problem in short time based on the traffic density in the opposite direction
62
SummarySummary
• Case Studies
• CASCADE-Based Advertising System• CASCADE-Based Merge Assistant System
• VANET Simulator
• Application-aware Simulator SWANS with Highway mobility (ASH)
Details are in the dissertation 63
Informational Applications
Entertainment Applications
SummarySummary
• Local View:– K. Ibrahim and M. C. Weigle. Accurate data aggregation for VANETs
(poster). In Proceedings of ACM VANET, pages 71-72, Montreal, Canada, Sept. 2007.
– K. Ibrahim, M. C. Weigle. Towards an Optimized and Secure CASCADE for Data Aggregation in VANETs (poster). In Proceedings of ACM VANET, pages 84-85, San Francisco, CA, Sept. 2008.
– K. Ibrahim and M. C. Weigle. Optimizing CASCADE data aggregation for VANETs. In Proceedings of the IEEE MoVeNet, pages 724-729, Atlanta, GA, Sept. 2008.
– K. Ibrahim and M. C. Weigle. CASCADE: Cluster-based accurate syntactic compression of aggregated data in VANETs. In Proceedings of IEEE AutoNet, New Orleans, LA, Dec. 2008.
SummarySummary
• Data Dissemination:– K. Ibrahim, M. C. Weigle. “p-IVG: Probabilistic Inter-Vehicle Geocast for
Dense Vehicular Networks”. In Proceedings of the IEEE VTC- Spring. Barcelona, Spain, Apr. 2009
• Security:– K. Ibrahim, M. C. Weigle. Securing CASCADE Data Aggregation for
VANETs. Poster in IEEE MoVeNet, Atlanta, GA, Sept. 2008.
– K. Ibrahim and M. C. Weigle. Light-weight laser-aided position verification for CASCADE. In Proceedings of the WAVE, Dearborn, MI, Dec. 2008.
• Simulation:– K. Ibrahim, M. C. Weigle. ASH: Application-aware SWANS with
Highway mobility. In Proceedings of IEEE MOVE, Phoenix, AZ, Apr. 2008.
Questions66
Thanks67