Multi-homed Vehicular Network Access Shreyasee Mukherjee with - - PowerPoint PPT Presentation

WINLAB

Multi-homed Vehicular Network Access

Shreyasee Mukherjee

with Akash Baid, Ivan Seskar and Dipankar Raychaudhuri WINLAB, Rutgers University, NJ

WINLAB

Outline

• Motivation
• Available Technologies
• Network Requirements
• The MobilityFirst Architecture
• Evaluation Use-cases
• Conclusion

2

WINLAB

Connected Vehicles

• Safety applications
• Demand for maps

and real-time traffic info

• Increasing demand

for infotainment based rich content delivery Motivation

3

WINLAB

Vehicular communication scenarios

• Vehicles with infrastructure (V2I/I2V)
• Vehicles with other vehicles (V2V)
• In-car communication

Focus

• n

V2I/I2V Focus

• n

V2I/I2V Motivation

WINLAB

4

WINLAB

• LTE
• WiFi

* www.att.com

AT&T LTE connection a Samsung Galaxy S4 Rutgers wireless on a Macbook Pro

* *

ǂ wigle.net

ǂ

Available technologies

Coverage Price Speed

Continuous coverage Good coverage but disconnection prone Expensive Inexpensive

5

WINLAB

Multi-homed Access

• How do we utilize

both of these technologies simultaneously?

• Network

efficiently utilizing multi-interface capability to improve user/network performance

WiFi LTE 6

WINLAB

Network requirements

• Accurate link

quality estimation

• Data striping

across multiple interfaces

• DTN mode of
• peration
• Seamless

handoff

7

WINLAB

The MobilityFirst Architecture

• Globally unique names

(GUIDs) for network attached objects

• Separation of names

from network addresses

• Storage-aware routing
• Hop by hop reliable

transport

8

WINLAB

Sender: GUIDX NA1 NA2 Re

• G

GNRS Server

Multi-homing in MobilityFirst

NCS GUID 11011..011 Send (GUID = 11011..011, SID=01, NA99, NA32, data) Send (GUID = 11011..011, data) GNRS update (GUID 11011..011, NA99, NA32, Stripe)

9

data striping uses a reliable backpressure based mechanism

WINLAB

Radio Parameter Value WiFi MAC 802.11a (non QoS) Propagation Log Distance Loss Rate Control Adaptive ARF LTE DL & UL RBs 15 MAC Scheduler Proportional Fair Tx power Dl-30dBm, UL- 10dBm

Dual-homing (WiFi + LTE) with DTN

• NS3 based simulations
• MobilityFirst naming routing

transport modeled

• NS3’s WiFi and LTE

modules used

• Multi-homed mobile client
• Stable LTE coverage with lower data

rate

• Intermittent WiFi coverage with

varying data rate

Evaluation

10

WINLAB

Throughput and delay metrics for dual-homing

10 20 30 40 50 60 70 80 90 20 40 60 80 100 120 140 160 180 200

Time(secs) Aggregate Throughput(MBytes) use both WiFi and LTE use only WiFi use only LTE 5meters/s(11mph) 10meters/s 20meters/s 30meters/s(67mph) 50 100 150 200 250 300 350 400 450 Speed of Vehicle File Transfer Completion Time (secs) use both WiFi and LTE use only WiFi use only LTE

No WiFi connectivity Aggregate throughput close to the sum of the throughput through each

Evaluation

11

WINLAB

Realistic (multi-WiFi +LTE) access scenario

• ~1X2 miles area of

San Francisco

• 30 open WiFi access

points shown

• Back-end LTE with

base-station at the approximate center assumed

• Actual cab traces in

the area simulated Evaluation

12

WINLAB

1 2 3 4 5 10 20 30 40 50 60 70

Average throughput per sec (in Mbps) Cab no.

1 2 3 4 5 10 20 30 40 50 60 70

Cab no. Maximum throughput per sec (in Mbps) Using only LTE Using the best available Wi-Fi Using all the available WiFis Using all the Wi-Fis and LTE

Multi-homing throughput for cabs

Average and maximum throughput for continuous data transfer of 100 seconds

Evaluation

13

WINLAB

Conclusion

• MobilityFirst features inherently supports multi-homing
• Future research directions:

Multi-network Access Probing for accurate link quality estimation Content server Content caching Improved Inter-domain Routing

Mobility prediction

Improved transport

14

WINLAB

Thank you!

WINLAB

Extras

16

WINLAB Planned WiFi hotspot deployment

17