Credits
• This material has been prepared by Giovanni Pau using papers and presentaGons from
InternaGonal Conferences and Journals
• The material publically available has been completed with original material.
Background
• What vehicles, bikes, busses, dolphins, zebra have in common?
• How they the items above are different from Cell-‐phones and what instead they have in common?
• What is the Major problem solved by TCP/IP and protocol Layering?
3
Background
• IntuiGvely what role play:
– Mobility?
– Physical Environment?
• What are the challenges?
A Taxonomy by Architecture
Wireless Networks
Infrastructure
Cellular WiMax Wifi Vanet (v2i) . . .
Peer-‐to-‐Peer
VaNet
(v2v) DTN Warfare . . .
5
Vehicular Networks
New Roles for Vehicles on the road
• Safe navigaGon:
– Vehicle & Vehicle, Vehicle & Roadway communicaGons
– Forward Collision Warning, Blind Spot Warning, IntersecGon Collision Warning…….
– In-‐Vehicle Advisories
• “Ice on bridge”, “CongesGon ahead”,….
7
New Roles for Vehicles (cont)
• Vehicle as content/entertainment sharing peers:
– Share locaGon criGcal mulGmedia files – Exchange local ad informaGon
– Support passenger to passenger internet games – Monitor PolluGon and opGmize traffic flow
– . . .
New Roles for Vehicles (cont)
• Vehicles as environment sensors:
– Pavement condiGon
– Probe data for traffic management – Weather data
– Physiological condiGon of passengers, ….
– Pervasive urban surveillance
– unconscious witnessing of accidents/
crimes
9
Convergence to a Standard:
Government, Industry, Academia
• Federal CommunicaGons Commission created DSRC
– … allocaGon of spectrum for DSRC based ITS applicaGons to increase traveler safety, reduce fuel consumpGon and polluGon, and
conGnue to advance the naGons economy.
• FCC Report and Order, October 22, 1999, FCC 99-‐305
• Amendment with licensing rules in December 2003
• DSRC Standards
– ASTM E17.51, IEEE 802.11p
– hmp://grouper.ieee.org/groups/scc32/dsrc/
• USDOT/CAMP have created CooperaGve IntersecGon Collision Avoidance (CICAS) ConsorGum
– hmp://www.its.dot.gov/cicas/cicas_workshop.htm
• AutomoGve companies created Vehicle Safety CommunicaGons ConsorGum (VSCC)
• Academia and Industry have sponsored several Special Issues, Workshops on the subject:
– VANET, V2VCom, Autonet, MoveNet, etc
The Standard: DSRC / IEEE 802.11p
• Car-Car communications at 5.9Ghz
• Derived from 802.11a
• three types of channels: Vehicle-Vehicle service, a Vehicle-Gateway service and a control broadcast channel .
• Ad hoc mode; and infrastructure mode
• 802.11p: IEEE Task Group for Car-Car communications
11 Forward radar
Computing platform Event data recorder (EDR)
Positioning system
Rear radar
Communication facility
Display
USDOT VII: Vehicle Infrastructure IntegraGon IniGaGve
• hmp://www.itsa.org/vii.html
– The VII IniGaGve is a cooperaGve effort between Federal and state departments of transportaGon (DOTs) and vehicle manufacturers to evaluate the technical, economic, and social/poliGcal feasibility of deploying a communicaGons
system to be used primarily for improving the safety and efficiency of the naGon's road
transportaGon system.
Vehicular Paradigms
• Vehicle to Vehicle (V2V): presents the
challenges typical of an ad hoc network in
addiGon to a very high speed mobility and an intermiment connecGvity
• Vehicle to Infrastructure (V2I): Protocol design is challenged by intermiment connecGvity and short communicaGon windows.
• OpportunisGc: V2V for a limited number of hops unGl is possible to connect to the
Infrastructure.
13
Mobility and PenetraGon raGo.
• AD Hoc networks
– Historically designed for specific applicaGons or user
groups. The network is deployed on demand in a specific condiGon within a specific group. The network scale is a generally not a main issue, the network is almost always connected.
• Vehicular Networks
– The network includes several thousand of nodes,
potenGally every car. Is not given that all the cars are connected or equipped and is important to understand the physical channel, and the physical scenario roles.
PenetraGon and deployment to be carefully designed.
– Scale could become an issue if a flat model is assumed.
General ConsideraGons
• New technologies and ships in
communicaGon paradigms are posing a
number of challenges to the current Internet architecture.
• The Wireless Challenges:
– Mobility
– Intermiment connecGvity (i.e. PropagaGon) – Long Delays (DTN)
– Path Discovery and RouGng – . . .
15
Challenge #1 Mobility
City secGon mobility model (Camp 02)
• RWP on general connected domain
courtesyof J.-Y.
LeBoudec, EPFL 17
RWP with locality (Blazevic 04)
• Stay in one
subdomain for some Gme then move to another
• Can also model city secGon mobility with locality
courtesyof J.-Y.
LeBoudec, EPFL
Example: Portland, Oregon (cont’d)
• Some staGstcs:
– Cars: 16,000/350
– Area: downtown Portland (5kmx7km) – Granularity 1sec
– Microscopic traffic simulaGon – Data: US Bureau of Census
– Maps: Tiger/ Portland transportaGon Authority
19
Trace Driven Mobility
• SynteGc: Fine grain vehicular traces are generated by using a microscopic traffic simulator (I.e. CORSIM, TRANSITSIM, etc)
– Issues:
• Map details (Tiger Database, signals, buildings, etc)
• AcGviGes (why the cars move?)
• Amount of data (15 minutes= about 4GB)
• Actual: Traces are collected from actual vehicle, and a model is inferred.
– Issues:
• Type of mobility
• GeneralizaGon of the model
Mobility Comparison
• Portland Oregon
• 5x7 KM
• Transims/NS2
21
Example: Portland, Oregon.
• CORSIM simulated traces starGng from the US bureau of census data.
• Tiger Database Maps, integrated with the street signal Gming.
• Access point locaGons in Portland downtown are used to perform connecGvity simulaGons.
• Detailed propagaGon models used to perform the simulaGons.
Actual Traces
• No real actual traces publically available yet!
• Available
– Bus Aggregated Traces
– Campus pedestrian traces (Crawdad) – Emergency Vehicle traces
23
VANET detailed simulaGons
Traffic Simulator
Maps
Signals
Light Gming AcGons
PopulaGon Stats
Mobility Traces
Buldings PropagaGon details
Network Simulator
Challenge #2: Physical Envionment
25 Paris, Oct '14
State of the Art
• SimulaGon is key to assess large scale performance; however very detailed propagaGon models are rare.
– NS2 à Two Ray + Shadowing in some cases – Qualnet à Two Ray, Ricean model and
Shadwoing, also has a number of Terrain models that take in consideraGon the Z factor.
• Missing:
– Urban PropagaGon model (i.e. buildings exist)
• Released by UCLA the corner model not perfect but good enough.
Why PropagaGon Mamers
Topology Two Ray Corner
27
CORNER: A Step Towards RealisGc SimulaGons for VANET – In proceedings of ACM VANET,
co-‐located with Mobicom 2010.
E. Giordano, R. Frank, G. Pau, M. Gerla
PropagaGon in Urban Scenarios
• VANET studies were
performed mainly using
“flat” propagaGon schemes (Two Ray)
• In reality propagaGon is affected by obstacles (buildings)
Paris, Oct '14
MoGvaGon
• Flat propagaGon schemes lead to unrealisGc results
• Ray Tracing techniques are too expensive:
– ComputaGon Gme
– 3D DescripGon of the environment
• Must find a trade-‐off between cost and realism
– Urban propagaGon Path Loss formulae
Path Loss Formulae
Analy<cal formulae for path loss predic<on in urban street grid microcellular environments,
Q Sun, SY Tan, KC Teh -‐ IEEE TransacGons on Vehicular Technology, 2005 -‐
ieeexplore.ieee.org
€
PL =10log λ 4π
$
% & ' ( )
2 λ
4rmrs2 + λ
4π
(
rm + rs)
$
%
&
&
' ( ) )
2
10
LW Nmin 10
* + , ,
- . / /
rm
rs
Tx
Jm
Rx
RW
m RW
s
31
CORNER at a glance
• Relate the posiGon of
nodes to the environment
– Specialized Reverse Geocoding
• Assess the propagaGon situaGon
– LOS – NLOS1 – NLOS2
• Compute formula parameters
• Apply the formula and compute the Path Loss
Tx
Rx
Jm rm
rs Wm
WS
CORNER: Reverse Geocoding
• Regular Reverse Geocoding does not apply
• Need to assess the signal
path. 1 A B
2
3
4
• Define a proximity area
• Assign each vehicle a set of possible road segment
• Find the pair of segments that involves the least
number of intersecGons (corners)
33
CORNER: Reverse Geocoding
• Regular Reverse Geocoding does not apply
• Need to assess the signal path.
• Define a proximity area
• Assign each vehicle a set of possible road segment
1 2
6
3 4 5
7 B
A
• Find the pair of segments that involves the least
number of intersecGons (corners)
Angular View
CORNER: SituaGon Assessment
• Same road
segment: LOS
• Adjacent road segment: either LOS or NLOS1
NLOS1 LOS
35
AWA AWB
CORNER: SituaGon Assessment
NLOS1 LOS
NLOS2
• Same road
segment: LOS
• Adjacent road segment: either LOS or NLOS1
• Connected road segment:LOS,
NLOS1 or NLOS2
Formula ApplicaGon
• Once propagaGon situaGon is assessed
• Formula parameters can be computed
• Apply the formula and obtain Path Loss
rm
rs
Tx Jm
Rx
RWm
D R RWs
37
PropagaGon Example
• Fixed Tx
• Mobile Rx
– Moves first from LOS to NLOS1 then to NLOS2
150m
150m 150m
Tx
1st Turn
2nd Turn
-160 -140 -120 -100 -80 -60 -40 -20 0
0 50 100 150 200 250 300 350 400 450
Path Loss [dB]
CORNER CORNER + Fading
-160 -140 -120 -100 -80 -60 -40 -20 0
Path Loss [dB]
CORNER
IntersecGons provide wider coverage
Source placed in the middle of the block Source placed at the intersecGon
Observa<on #1: Vehicles at intersec<ons can cover a much wider area
39
CORNER: ValidaGon
• Instrumented 2 cars with:
– A laptop
– A IEEE802.11b/g wireless card – GPS receiver
• Each car broadcasts its posiGon
– Using raw sockets, directly at layer 2
– 10 Gmes per second
• 3 experiment sets
– Fixed to mobile – Mobile to mobile – Fixed to fixed
Fixed to Mobile
• The fixed node is periodically
broadcasGng
• The mobile node saves the locaGon where it receives the packet
• Two nodes
– One Fixed
– One revolving around the block
• Comparison with simulaGon
41
Mobile to Mobile
• Two cars revolving around the block in opposite
direcGon
• Each one broadcasts its own posiGon
• A link is plomed from the point the packet was sent to the point the packet
was received
Field Experiment
Fixed to Fixed
• Fixed sender
• Fixed receiver in different posiGons
• Compute the link quality as:
– # Packets Received / # Packets Sent
0 0.2 0.4 0.6 0.8 1
0 1 2 3 4 5
Link Quality
Field Experiment CORNER + Rice CORNER + Rayleigh
43
CORNER: Impact (1)
Range 80m Range 250m CORNER
Range 80m Range 250m CORNER
Node Degree Average 7.6 47.2 19.9
Link DuraGon: Average 11.6 24.52 16.2
ConnecGvity Index: 0.69 0.99 0.99
Average Number of Hops: 4.2 1.5 2.3
CORNER: Impact (2)
• 80m range too pessimisGc
• 250m range too opGmisGc
• There might be a range that approximates bemer, but depends on the map.
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0 5 10 15 20 25
Packet Delivery Ratio
% of Transmitting Nodes
CORNER Two Ray 250m Two Ray 80m
45
Remarks
• CORNER
– A good trade off between realism and computaGonal cost
– Could be implemented for any network simulator – Independent of the mobility simulator
• hmp://nrl.cs.ucla.edu/~egiordano/vergiliusJoomla/
– Source code of QualNet implementaGon
– Binaries for the generaGon of Path Loss Matrix
VERGILIUS: a Scenario Generator for VANET
E. Giordano, E. De Sena, G. Pau, M.
Gerla
hmp://cvet.cs.ucla.edu/vergilius.html
Paris, Oct '14 47
IntroducGon
• For VANET simulaGon is essenGal:
– Large scale test beds are impossible
• Crucial that simulaGons reflect reality
– Mobility
• Real traces: realisGc but too specific
• SyntheGc traces: not as realisGc but tunable
– PropagaGon
• A propagaGon model that reflects reality is needed
IntroducGon (2)
• VERGILIUS implements:
– A tunable scenario generator:
• Provides realism when needed
• Easily tunable traffic characterisGcs
– A realisGc PropagaGon Model: CORNER
• Takes into account the presence of buildings
• Light weight computaGon, can be run on the fly.
– An extensive trace analyzer
• Extract characterisGcs of the mobility
49
Sample Workflow
QualNet
CORSIM TIGER-‐CENSUS
Geographic Database:
provides the road topology
Mobility Simulator:
provides detailed road traffic traces
Network Simulator:
provides network metrics Scenario Generator:
Provides control over the kind of traffic to generate (uniform, aggregated etc.)
Trace Analyzer:
Provides:
– PropagaGon matrix considering buildings.
– ConnecGvity and Interference metrics based on topology
Scenario Generator
Mobility Simulator
Mobility Trace Map
Database
Map DescripGon
• Nodes
• Edges
Scenario
• Map
• Traffic VERGILIUS
• Traffic Scenario:
– Input Flows & Turn Probabilites – Route DescripGons
User Input
Equivalent
51
Routes GeneraGon
• All routes go from an entry point to an exit point
• The number of routes originaGng from one entry point can be set
– DeterminisGcally – Randomly
• Uniform DistribuGon
• Poisson DistribuGon
• Based on the importance of the road
• The desGnaGon for each route can be set:
– DeterminisGcally – Randomly
• The Route will take the Dijkstra shortest path based on:
– Distance
– Time to Traverse
Choice of Route DesGnaGon
• Random tunable distribuGon
• Trip AggregaGon Factor (TAF)
• Using TAF we can change the nature of the traffic.
TAF = 0 TAF = 5 TAF = 100 53
Effect of the Trip AggregaGon Factor
SCENARIO RANDOM
SCENARIO_DIJKSTRA_WEIGHT TIME TAF = 4
AVERAGE_ARRIVALS 4200
ENTRY_FLOW_MODE WEIGHTED SCENARIO TOPOLOGY_BASED
ENTRY_FLOW_MODE UNIFORM TB_INPUT_FLOW CONSTANT 360 SCENARIO_DIJKSTRA_WEIGHT TIME
PropagaGon: IntroducGon
• VANET studies were
performed mainly using “flat”
propagaGon schemes (Two Ray)
• In reality propagaGon is affected by obstacles (buildings)
Paris, Oct '14
The PropagaGon Model: CORNER
• Place vehicles on road segments (reverse
geocoding)
• Two vehicles:
– Same road segment: LOS – Adjacent road segment:
either LOS or NLOS1
– Connected road segment:
– LOS, NLOS1 or NLOS2
Sight Window
for second crossroad NLOS1
LOS NLOS2 Sight Window for first crossroad Sight Window
NLOS1 LOS
LOS, NLOS1, NLOS2 amenuaGon formulae provided by:
Analy<cal formulae for path loss predic<on in urban street grid microcellular environments, Q Sun, SY Tan, KC Teh -‐ IEEE TransacGons on Vehicular Technology, 2005 -‐ ieeexplore.ieee.org
CORNER: Example
• Fixed Tx
• Mobile Rx
– Moves first from LOS to NLOS1then to NLOS2
150m
150m 150m
Tx 1st Turn
2nd Turn
-160 -140 -120 -100 -80 -60 -40 -20 0
0 50 100 150 200 250 300 350 400 450
Path Loss [dB]
Manhattan Distance [m]
CORNER CORNER + Fading
57
Trace Analyzer (1)
Scenario A Scenario B
Average Number of Nodes: 91.78 92.61
Node Degree: Average / Variance 13.16 / 121.37 24.42 / 234.72 Link DuraGon: Average / Variance 3.95 / 39.98 11.15 / 256.19
ConnecGvity Index: 0.96 0.96
Average Number of Disconnected Nodes: 0.61 0.42
Average Number of Hops: 2.91 2.25
Average Hidden Nodes: 874.22 1828.25
• Average Number of Nodes: Number of nodes present in the area averaged over Gme.
• Node Degree: Number of neighbors of each node, averaged over each node and over Gme
• Link Dura<on: DuraGon of the connecGon between a pair of nodes averaged over all pairs and over Gme
Trace Analyzer (2)
Scenario A Scenario B
Average Number of Nodes: 91.78 92.61
Node Degree: Average / Variance 13.16 / 121.37 24.42 / 234.72 Link DuraGon: Average / Variance 3.95 / 39.98 11.15 / 256.19
ConnecGvity Index: 0.96 0.96
Average Number of Disconnected Nodes: 0.61 0.42
Average Number of Hops: 2.91 2.25
Average Hidden Nodes: 874.22 1828.25
• Connec<vity Index (CI): defined as the average porGon of the
network that is reachable from each vehicle in the mobility trace, regardless of the path length. CI provides a bemer insight onto the parGGoning of the network.
59
Trace Analyzer (3)
Scenario A Scenario B
Average Number of Nodes: 91.78 92.61
Node Degree: Average / Variance 13.16 / 121.37 24.42 / 234.72 Link DuraGon: Average / Variance 3.95 / 39.98 11.15 / 256.19
ConnecGvity Index: 0.96 0.96
Average Number of Disconnected Nodes: 0.61 0.42
Average Number of Hops: 2.91 2.25
Average Hidden Nodes: 874.22 1828.25
• Average Number of Disconnected Nodes (AD): the average number of nodes that are disconnected from the network, i.e. that have no neighbors. It provides informaGon on how open there are vehicles that are separated from the rest of the network.
• Average Number of Hops (AH): the average number of hops needed to reach all nodes in the network.
Trace Analyzer (4)
Scenario A Scenario B
Average Number of Nodes: 91.78 92.61
Node Degree: Average / Variance 13.16 / 121.37 24.42 / 234.72 Link DuraGon: Average / Variance 3.95 / 39.98 11.15 / 256.19
ConnecGvity Index: 0.96 0.96
Average Number of Disconnected Nodes: 0.61 0.42
Average Number of Hops: 2.91 2.25
Average Hidden Nodes: 874.22 1828.25
• Average Hidden Nodes: the average number of nodes that have two neighbors that are not each other neighbors.
A
C B
61
Trace Analyzer (5)
• Hops Distribu<on Func<on (HDF): the average porGon of
network that is reachable given a maximum number of hops.
Provide informaGon about the stretch of a network, i.e.
what is the average distance in hops needed to traverse the network.
0 0.2 0.4 0.6 0.8 1
1 2 3 4 5 6 7 8 9 10 11 Por<on of Network Reachable
Number of Hops
Scenario A Scenario B
Challenges and OpportuniGes in Urban Vehicular Systems
An Analysis from a RealisGc Trace (Eugenio Giordano, Giovanni Pau,
Antony Rowstron)
Paris, Oct '14 63
DescripGon of The Trace
• 15 minutes 1 second granularity
• Portland downtown @ 8am –8.15am
• 16000 different vehicles
• Avg of 3500 vehicles in the trace
• SyntheGc trace developed by the Los Alamos NaGonal Lab for naGonal security.
• Based on AcGvity surveys
Vehicles density
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0.0 4.2 8.5 12.7 17.0 21.2 25.5 29.7 34.0 38.2 42.5 46.7 51.0 55.2 59.5 63.7 68.0 72.2 76.5 80.7 85.0 89.2 93.5 97.7 102.0 106.2 110.5
Cumula<ve Por<on of Vehicles
Distance From The Center of The Closest Intersec<on [m]
Distance From Intersec<ons
Observa<on #2: Vehicle density is higher at intersec<ons
50% of the vehicles is within 25 meters of an intersecGon
65
ParGGons Analysis
• Is the network really parGGoned?
– The median of the parGGon size is 1, this
suggests that there are many small parGGons and a few large parGGons. I believe this needs some further invesGgaGon.
– ConnecGvity Index (CI) Analysis
• CI is the number of vehicles a vehicle can reach regardless of the number of hops
• Compute the average CI per second then CDF or Average… Let’s see the results
Reachability
• Reachability: porGon of the network that a node can reach.
• How to read the cumulaGve: For reachability X the porGon of network that can reach at most the porGon X of the network
• Two different results:
– For low penetraGon raGos the network is parGGoned. E.G. for 0.0625 we have that all network has at most 0.51 reachability (meaning no node can reach more than 51% of the network)
– For higher penetraGon raGos the network is not parGGoned. A small porGon of nodes is isolated, and the rest of the network is almost fully connected.
0 0.2 0.4 0.6 0.8 1 1.2
0 0.03 0.06 0.09 0.12 0.15 0.18 0.21 0.24 0.27 0.3 0.33 0.36 0.39 0.42 0.45 0.48 0.51 0.54 0.57 0.6 0.63 0.66 0.69 0.72 0.75 0.78 0.81 0.84 0.87 0.9 0.93 0.96 0.99
Cumula<ve Por<on of Network
Reachability
0.00390625 0.0078125 0.015625 0.03125 0.0625 0.125 0.25 0.5
67
ConnecGvity Index (CI)
• CI is the average of the reachability.
– Reflects the result of previous graph
– At PenetraGon R 0.125 the network connects
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0.00390625 0.0078125 0.015625 0.03125 0.0625 0.125 0.25 0.5 1
Connec<vity index
Penetra<on Ra<o
Connec<vity Index:
Hop Distance Analysis
• How far are the nodes in terms of hops?
– As the PenetraGon R increases each node can reach more porGon of the network, the stretch increases
– When the network connects (0.125) we have a 10 hops distance (on average!!!)
• Is a flat mulGhop network feasible?
– How many hops away can 2 nodes communicate?
0 5 10 15 20 25 30
0.00390625 0.0078125 0.015625 0.03125 0.0625 0.125 0.25 0.5 1
Average Distance [Hops]
Penetra<on Ra<o
Average Number of Hops: Median Number of Hops: 99th PercenGle Number of hops
69
SoluGons
• The Network is parGGoned and too big:
– SoluGon:
• DisrupGon/Delay tolerant
§ +No Management
§ +Comes at no cost
§ -‐ Delay (probably not for human interacGon)
• Infrastructure Aided
§ -‐ Cost
§ -‐Management
§ +Shorter paths, more coverage -‐> low delay
3 AP placement methods:
• AP are ALWAYS placed at intersecGons
• IMPORTANCE: based on the density of the
• GRID: IntersecGons closest to the ideal grid
• Random: IntersecGons chosen Randomly
• AP ARE NEVER on FREEWAYS.
71
IMPORTANCE GRID RANDOM
Methods Comparison (Penetr R = 1)
0 0.2 0.4 0.6 0.8 1 1.2
0.1 0.2 0.5 1 2 5 10
Network Covered Within 3 Hops
AP Density [APS/ Km2]
importance grid
Random
0 0.2 0.4 0.6 0.8 1 1.2
0.1 0.2 0.5 1 2 5 10
Por<on of Nework Within 5 Hops
AP Density [APs / Km2]
importance grid
Random
73
Infrastructure Hops distance vs AP density
AP placement: GRID
• Full connecGvity at 1 hop (V2I only) is not feasible
• With 10 APs/Km2 the whole network is within 3 hops of the infrastructure
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0.1 0.2 0.5 1 2 5 10 15 20 25 30 40 50 75 100
Por<on of networ Reachable
Infrastructure Density [APs/Km2]
Network Reachable vs Infrastructure Density
1 hop 3 hops 5 hops 2 hops
Infrastructure Benefits
• DistribuGon of Distance from
infrastructure for PR = 1 and the AP placement based on the density of nodes.
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
Por<on of Network
Distance from Infrastructure [Hops]
AP Placement: Importance Penetra<on R = 1
"Density 0.1 AP/Km2"
Density 0.2 AP/Km2 Density 0.5 AP/Km2 Density 1 AP/Km2 Density 2 AP/Km2 Density 5 AP/Km2 Density 10 AP/Km2
75
Infrastructure aided (cons)
• It needs management!!
– DHT or GHT
– Evaluate the overhead cost in terms of Number of updates needed
Infrastructure aided: cost
• Updates increase
• AssociaGon Gme decreases
0 0.02 0.04 0.06 0.08
0.1 0.2 0.5 1 2 5 10
Average Updates per second per vehicle
AP Density [APs /Km2]]
Average Updates per second per vehicle
DENSITY GRID
IMPORTANCE RANDOM
0 20 40 60 80 100
0.1 0.2 0.5 1 2 5 10
Median inter-‐update Time [s]
AP Density [APs /Km2]]
Median inter-‐update Time [s]
DENSITY GRID
IMPORTANCE RANDOM
77
Challenge #3 ApplicaGon Design
Issues
• Mobility and PropagaGon result in
disconnecGons and large delays, applicaGons need to be a-‐synchronous to cope with such events.
• Few Examples in following slides.
79
Used Tools
QualNet
CORSIM TIGER-‐CENSUS
Geographic Database:
provides the road topology
Mobility Simulator:
provides detailed road traffic traces
Network Simulator:
provides network metrics Scenario Generator:
Provides control over the kind of traffic to generate (uniform, aggregated etc.)
Trace Analyzer:
Provides:
– PropagaGon matrix considering buildings.
– ConnecGvity and Interference metrics based on topology
hmp://cvet.cs.ucla.edu/vergilius.html
CarTorrent : Opportunistic Ad Hoc networking to download large
multimedia files
Alok Nandan, Shirshanka Das Giovanni Pau, Mario Gerla
WONS 2005
81 Paris, Oct '14
You are driving to Vegas
You hear of this new show on the radio Video preview on the web (10MB)
One option: Highway Infostation download
Internet
file
83
Incentive for opportunistic “ad hoc networking”
Problems:
Stopping at gas station for full download is a nuisance Downloading from GPRS/3G too slow and quite
expensive
Observation: many other drivers are interested in download sharing (like in the Internet)
Solution: Co-operative P2P Downloading via Car-Torrent
CarTorrent: Basic Idea
Download a piece
Internet
Transferring Piece of File from Gateway Outside Range of Gateway
85
Co-‐operaGve Download: Car Torrent
Vehicle-‐Vehicle Communica7on
Internet
Exchanging Pieces of File Later
BitTorrent: Internet P2P file downloading
Uploader/downloader
Uploader/downloader
Uploader/downloader
Uploader/downloader
Tracker Uploader/downloader
87
CarTorrent: Gossip protocol
A Gossip message containing Torrent ID, Chunk list and Timestamp is “propagated” by each peer
Problem: how to select the peer for downloading
SelecGon Strategy CriGcal
89
