Transcript Apsitti99 - Tohoku University

Monitoring vehicles
The potential and the challenge
Kohei OHTA
Introduction

Connect CARs with Internet
• Seamless connectivity
• Connect to World
• Connect to various personal equipments
• Many new technologies available
• MIPv6
• Auto configuration
• Ad-hoc networking
• Many NEW mobile applications for CARs
• Road service
• On-line customer service
2004/1/28
2
Probe Car project

Develop new applications with car related
information like
•
•
•
•
•
2004/1/28
Velocity, Acceleration
GPS information
Winker status
Wiper status
Etc…
3
Requirements for information
collection from Cars

Security
• Protect privacy information

Standardization
• Naming
• Common scheme and methods the various types of
information
• Transport protocol
• Tough transport protocol under wireless environment

Efficiency
• High resolution and row bandwidth transport
2004/1/28
4
Security

There are many privacy sensitive information
in cars
•
•
•
•

Geographical location
Velocity
Passengers
Etc
Strong security mechanism is required
• Encryption
• Authentication
• Access control
2004/1/28
5
Standardization

Naming
• To support multi vendor environment, common
scheme to specify a information item.
• For example, we have to be able to access vehicle
velocity from any vendor’s car with SAME method.

Transport
• Simple and tough transport protocol is suitable for
mobile wireless environment
2004/1/28
6
Efficiency

Mobile wireless environment is worse than
wired network
• Connectivity is NOT guaranteed at all
• There is less chance of stable connectivity
• The bandwidth is often narrow

It is difficult to collect high resolution
information in such environment
• UDP is preferable to TCP
• Effective aggregation for high resolution with less
bandwidth under unreliable line
2004/1/28
7
Applicability of SNMPv3 [1]

Information collection in TCP/IP is
• Basic issue for network management

That is already existing
• There are many management information
for various networked devices and
applications, router, switch, server, printer,
and so on

Well standardized
2004/1/28
8
SNMPv3 has

For Privacy
•

For authentication
•
•

Different key management with single pass phrase
For access control
•

Multiple user support
MD5/SHA1 digital signature
Key management
•

DES/AES encryption support
Allow access to iitsVehicleVelocity from center A
Deny access to iitsLongitude from service center B
Deny access to iitsLatude from service center B
Strandardization
•
Well standardized consistent naming scheme
• .1.3.6.1.4.1.282.16.108.1.3.3.3 iitsVehicleVelocity
•
Transport
• Connectionless UDP
• Per packet authentication/encryption
2004/1/28
9
Current configurations
Information collection center
Encrypted
IPv6
Information collection center
Velocity
Logitude
Latitude
Defined as MIB in ASN.1
Winker
Wiper status
2004/1/28
10
Polling and Trap

There are two modes in SNMP to collect information
• Polling is good to collect periodic data
• Trap is good to notify a event
2004/1/28
11
For effectiveness
SNMP Packet Payload
Manager
Get
Manager
Response
Agent
Name-1
null
null
Agent
Name-1 Value-1
1.3.6.4.1.282.16.107.2.1.1.3
2004/1/28
Name-n
Name-n Value-n
646213932
12
Polling at small intervals
Manager
Manager
time
Agent
Query and response
60 MOs/packet
Query and response
Current Method
time
Agent
Delay
Responze
Name-1
Value-1
Name-2
Value-2
Name-(n-1)
Value-n
PDU SIZE = 60 * (length (Name) + length(Value))
Length(Name) = 12;
length (Value) = 6
= 60 * (12 + 6 ) = 1.08K bytes !!
2004/1/28
13
Time-based (History) Aggregation MO
Manager
MO: n values of ifInOctets.１
interval I
Agent
Get
Name-A
Manager
null
Agent
Response
Name-A
Value-A
t1,Val(t1),Val(t1+I),Val(t1+2I),….Val(t1＋(n-1)I)
Responze
Name-A
t1
Val(t1) Val（ｔ１＋I）
Vaｌ（t1+(n-2)I) Vaｌ（t1+(n-1)I
PDU SIZE ~ 60 * length(Value))
length (Value) = 6
~ 60 * 6 = 360 bytes !!
2004/1/28
14
Evaluation - Data Retrieval Time
2004/1/28
15
Evaluation – Management Traffic
2004/1/28
16
Evaluation - CPU Usage
2004/1/28
17
For reliable information collection




Data collection with SNMP has been based
on Polling
Mobile node is often disappeared from data
collector because of the physical condition
and occasional shadowing.
Data collector can not know and control the
connectivity to mobile node.
Conventional polling based data collection no
longer work well.
2004/1/28
18