Now you can track up to 690 meters.
Maybe soon online advertising will literally track you down. A new way to find out where you are by watching your internet connection can set your current location to within a few hundred meters.
Such techniques are already used, but they are much less accurate. Each computer connected to the web has an IP address, but there is no easy way to map it to a physical location. The best of the existing systems can give an error of up to 35 kilometers. (Probably referring to GeoIP - approx. Transl.)
Now, to achieve much greater accuracy, Yong Wang, from the University of Electronics and Technology of China in Chengdu, and his colleagues from Northwestern University of Evanston, Illinois, used various companies and universities as reference points.
Such organizations often place their sites on servers located in their own buildings; This means that their IP addresses are tied to the physical location. To determine the physical and web addresses of such organizations, the Wan group used Google Maps; and thus gained about 76,000 landmarks. For comparison, the majority of other geolocation methods use several hundred landmarks, and they are specially placed for this purpose.
')
The new method finds the target computer in three steps. At the first stage, the time required for the data packet to reach the target is measured, and converted into a distance — a common technique in geolocation — it narrows the possible location of the target to an area of about 200 km.
Then Van and his colleagues send packets to well-known reference servers (from Google Maps) located in this area in order to determine which routes they follow. When the landmark and target host are behind a single router, researchers can compare how much a packet travels from the router to each of the hosts; this time difference, converted into an approximate difference in distance, narrows the search even more. “We are reducing the size of the area in which our target is potentially located,” explains Van.
Finally, they repeat this search by reference points at a more detailed level: once again comparing the time delays, they establish which of the reference servers is the closest to the target. Of course, the result may not be completely accurate, but it is much better than trying to determine a place by simply recounting the delay from the first stage to the distance; and better than the best of the existing methods based on IP address. On average, their method gives a variation of 690 meters, and at best it reaches 100 meters - enough to locate the target computer within several streets.
This accuracy class usually requires a person to consciously disclose his location (probably meant launching client software, such as in smartphones - approx. Transl.) , But the Vaughn method works without the user's consent. As he put it, “This method is customer-independent. The client does not need to confirm anything. ”
You can completely avoid geolocation by sending your traffic through a proxy server; then it will seem that you are in a different place. Vaughn can't get around this; but says that you can at least determine the presence of a proxy and return an empty result instead of a false one.
Although the Vaughn method will potentially allow targeting of advertisements right down to a specific street, advertisers themselves may prefer to maintain a wider targeting. “The bulk of brands will not necessarily want such selectivity,” comments Jack Wallington of the London Internet Advertising Bureau . He notes that the method, however, may be useful in some situations; for example, for hungry office workers, you can show the nearest fast food coupons.
Last week, Vaughn presented his research at the Usenix Symposium on Networked Systems Design and Implementation in Boston.
Such techniques are already used, but they are much less accurate. Each computer connected to the web has an IP address, but there is no easy way to map it to a physical location. The best of the existing systems can give an error of up to 35 kilometers. (Probably referring to GeoIP - approx. Transl.)
Now, to achieve much greater accuracy, Yong Wang, from the University of Electronics and Technology of China in Chengdu, and his colleagues from Northwestern University of Evanston, Illinois, used various companies and universities as reference points.
Such organizations often place their sites on servers located in their own buildings; This means that their IP addresses are tied to the physical location. To determine the physical and web addresses of such organizations, the Wan group used Google Maps; and thus gained about 76,000 landmarks. For comparison, the majority of other geolocation methods use several hundred landmarks, and they are specially placed for this purpose.
')
Approaching
The new method finds the target computer in three steps. At the first stage, the time required for the data packet to reach the target is measured, and converted into a distance — a common technique in geolocation — it narrows the possible location of the target to an area of about 200 km.
Then Van and his colleagues send packets to well-known reference servers (from Google Maps) located in this area in order to determine which routes they follow. When the landmark and target host are behind a single router, researchers can compare how much a packet travels from the router to each of the hosts; this time difference, converted into an approximate difference in distance, narrows the search even more. “We are reducing the size of the area in which our target is potentially located,” explains Van.
Finally, they repeat this search by reference points at a more detailed level: once again comparing the time delays, they establish which of the reference servers is the closest to the target. Of course, the result may not be completely accurate, but it is much better than trying to determine a place by simply recounting the delay from the first stage to the distance; and better than the best of the existing methods based on IP address. On average, their method gives a variation of 690 meters, and at best it reaches 100 meters - enough to locate the target computer within several streets.
We do not depend on the client
This accuracy class usually requires a person to consciously disclose his location (probably meant launching client software, such as in smartphones - approx. Transl.) , But the Vaughn method works without the user's consent. As he put it, “This method is customer-independent. The client does not need to confirm anything. ”
You can completely avoid geolocation by sending your traffic through a proxy server; then it will seem that you are in a different place. Vaughn can't get around this; but says that you can at least determine the presence of a proxy and return an empty result instead of a false one.
Although the Vaughn method will potentially allow targeting of advertisements right down to a specific street, advertisers themselves may prefer to maintain a wider targeting. “The bulk of brands will not necessarily want such selectivity,” comments Jack Wallington of the London Internet Advertising Bureau . He notes that the method, however, may be useful in some situations; for example, for hungry office workers, you can show the nearest fast food coupons.
Last week, Vaughn presented his research at the Usenix Symposium on Networked Systems Design and Implementation in Boston.
Source: https://habr.com/ru/post/117151/
All Articles