| The rapid growth of number of Internet users has brought new opportunities for the development of Internet services, accompanied by new challenges. Knowledge of locations of users is needed in many functions such as user recommendations and directional advertisement in many Web services and to determine a user’s location can bring direct economic benefits. Besides, IP geolocation technology is meaningful in areas such as security. Existing IP geolocation services based on geoIP databases are unable to satisfy business applications for disadvantages in precision and coverage.After deep research and analysis of existing IP geolocation models based on network measurements and trends in them, a model of two levels that is able to locate an IP address in city is designed, considering characters of China’s network. The main work is as follows:(1) Based on the research and analysis of other IP goelocation models based on network measurements, accompanied by the research of China’s network, a two-level model is designed.(2) In the first level of locating an IP address to a city, topology based method is taken advantage of. IP blocks from BGP tables are partitioned according to nearest-block rule. IP addresses of routers are clustered in an iterative way and the geoIP database is updated using the results.(3) In the second level of locating an IP address in a city, the delay-based method is taken advantage of. Algorithms are used to solve problems of router path differing in different measurements and router inflation. The delays between targets and landmarks are estimated by traceroute program and are used to locate the targets with the model named Closed-Shortest.It’s proved in experiments that it is effective to update the geoIP database with network topology. A result with a quite small median error distance is acquired in the process of locating IP addresses in a city. As a whole, the two-level model makes great advances compared to other jobs done before in mainland. |
PDF Full Text Request