assessing the geographic resolution of exhaustive tabulation for...

Vermelding onderdeel organisatie

1

Assessing the geographic resolution ofexhaustive tabulation for geolocating

Internet hosts

S. Siwpersad, S. Uhlig

Delft University of TechnologyThe Netherlands

B. Gueye

Universite de LiegeBelgium

OUTLINE 2

Agenda

• Introduction• Database-driven IP geolocation• Measurement-based IP geolocation• Geographic resolution of databases• Conclusions

PART I - INTRODUCTION 3

Introduction

• GPS-based geolocation• Satellite information• Exact location

• IP geolocation• Round trip time• Inferred location

OUTLINE 4

Agenda

• Introduction• Database-driven IP geolocation• Measurement-based IP geolocation• Geographic resolution of databases• Conclusions

PART II - DATABASES 5

Structure of databases

• IP blocks• Geographic location names• Geographic coordinates• Additional information

• Example of database records:

PART II - DATABASES 6

Information sources for databases

• DNS:• Whois:

• User submitted information

Commercial geolocation databases

• Maxmind• 110 thousand cities• 3 million IP blocks (74% city-level, 4% country-level)

• 2 billion IP addresses

• Hexasoft• 15 thousand cities• 5 million IP blocks (67% city-level, 33% country-level)

• 4 billion IP addresses

PART II - DATABASES 7

Span of a city- or country name

PART II - DATABASES 8

DELFTTO

ROTTERDAM

AMSTERDAMTO

MILAN

LOSANGELES

TONEW YORK

Size of the largest cities

PART II - DATABASES 9

NEW YORK

BERLIN

ROTTERDAM

DUBLIN

Size of countries

PART II - DATABASES 10

NETHERLANDS

TURKEY

VATICAN CITY

USA

Differences between the databases

PART II -DATABASES 11ROTTERDAM

TOANTWERP

LOSANGELES

TONEW YORK

Drawbacks of databases

• Staleness of the location information• Incompleteness of the records within databases• Uncertainty on the used sources and methodology• Coarse granularity of the region• Discrete solution space

PART II - DATABASES 12

Advantages of databases

• Easy deployment• Fast lookups• Resource inexpensive• Suited for typical applications• Reasonably priced

PART II - DATABASES 13

OUTLINE 14

Agenda

• Introduction• Database-driven IP geolocation• Measurement-based IP geolocation• Geographic resolution of databases• Conclusions

Constraint-based geolocation

PART III - MEASUREMENTS 15

Area of the confidence region

PART III - MEASUREMENTS 16

MALTA PORTUGAL USA

Limitations of measurements

• Relies on replies from probing• Costly in terms of time and network resources• Unwanted distortions inherent to measurements

PART III - MEASUREMENTS 17

Advantages of measurements

• Up-to-date location information• Higher accuracy and finer granularity• Confidence on area or location estimate• Continuous solution space

PART III - MEASUREMENTS 18

OUTLINE 19

Presentation outline

• Introduction• Database-driven IP geolocation• Measurement-based IP geolocation• Geographic resolution of databases• Conclusions

Absolute resolution of databases

PART IV – RESOLUTION 20

WITHIN CONFIDENCE REGION OUTSIDE CONFIDENCE REGION

Relative resolution of databases

PART IV – RESOLUTION 21

OUTSIDE CONFIDENCE REGIONWITHIN CONFIDENCE REGION

OUTLINE 22

Presentation Outline

• Introduction• Database-driven IP geolocation• Measurement-based IP geolocation• Investigation on the resolution of databases• Conclusions

Conclusions

• Databases need improvements:• more complete records• meta-data about methodology

• Measurements are not always possible, but desirablefor better confidence, precision and validation

• Resolution of databases with respect to CBG is poor

PART V - CONCLUSION 23

PART V - CONCLUSION 24

Future work

• Investigate the quality of the information used fordatabases

• Add information in databases records (e.g.confidence)

• Automated IP geolocation method selection (active vs.passive)