http://repository.iiitd.edu.in/xmlui/handle/123456789/387 Sat, 11 Apr 2026 03:05:17 GMT 2026-04-11T03:05:17Z http://repository.iiitd.edu.in/xmlui/handle/123456789/442 Qi points : placing decoy routers in the internet Gosain, Devashish; Agarwal, Anshika; Acharya, Hrishikesh Bhatt; Chakravarty, Sambuddho Decoy Routing, the use of routers (rather than end hosts) as proxies, is a new direction in anti-censorship research. However, existing proposals require control of hundreds of Autonomous Systems (AS) to provide Decoy Routing to Internet users in a single censorious country (e.g. China). This is considered necessary, as the adversary - in this case the Chinese Government - has connections to many Autonomous Systems (ASes), and we want to make sure it cannot simply route around those ASes which have decoy routers. In this paper, we present a new approach to the question of placing decoy routers. In decoy routing, the router intercepts messages en route to an overt destination and proxies them to covert destinations. Instead of trying to capture flows from an entire country, as proposed, we stipulate that the overt destination be a well known site (such as Alexa top-100), and concentrate on the AS-level paths to these sites. We construct a map of the structure of the Internet, as a graph of such AS-level paths and present a new way to identify key points - those few ASes which appear on a large fraction of paths leading to these popular websites. Our method yields results an order of magnitude cheaper than earlier proposals, and needs to be run only once, rather than for each censorious country. (We also identify the key routers inside a few key ASes.) Our results indicate that decoy routing is much more powerful than previously believed: using our new approach to place decoy routers, we need very few (less than 0:1% of Internet AS) to force an adversary to route through them. However, while the number of key ASes is small, the number of key routers in these ASes may be quite large – a new challenge for decoy routing. Fri, 23 Sep 2016 04:51:19 GMT http://repository.iiitd.edu.in/xmlui/handle/123456789/442 2016-09-23T04:51:19Z http://repository.iiitd.edu.in/xmlui/handle/123456789/441 Cairn : identifying network locations for large scale censorship by resource-constrained adversaries Agarwal, Anshika; Gosain, Devashish; Acharya, Hrishikesh Bhatt; Chakravarty, Sambuddho Censorship of the Internet by government is a hotly contested topic. Some nations lean more toward free speech; others are much more conservative. How feasible is it for a government to censor the Internet? What mechanisms can it use? Where all should it install the censorship infrastructure? What collateral damage can be seen in other countries? In this paper, we attempt to look at these questions in general, and present a case study of India - a country which currently performs limited censorship, but which will likely change its access policies in the near future. Fri, 23 Sep 2016 04:46:37 GMT http://repository.iiitd.edu.in/xmlui/handle/123456789/441 2016-09-23T04:46:37Z http://repository.iiitd.edu.in/xmlui/handle/123456789/388 Geographical bias in GitHub : perceptions and reality Rastogi, Ayushi; Nagappan, Nachiappan; Gousios, Georgios Open source development has often been considered to be a level playing field for all developers. But there has been little work to investigate if bias plays any role in getting contributions accepted. The work presented in this study tries to understand the influence of geographical location on the evaluation of pull requests in GitHub - one of the primary open source development platforms. Using a mixed-methods approach that analyzes 70,000+ pull requests and 2,500+ survey responses, we find that geographical location explains statistically significant differences in pull request acceptance decisions. Compared to the United States, submitters from United Kingdom (22%), Canada (25%), Japan (40%), Netherlands (43%), and Switzerland (58%) have higher chances of getting their pull requests accepted. However, submitters from Germany (15%), Brazil (17%), China (24%), and Italy (19%) have lower chances of getting their pull requests accepted compared to the United States. The chances of pull request acceptance decisions increase by 19% when the submitter and integrator are from the same geographical location. Survey responses from submitters indicate the perception of bias is strong in Brazil and Italy matching our results. Also, 8 out of every 10 integrators feel that it is easy to work with submitters from the same geographical location. Mon, 11 Jan 2016 11:54:45 GMT http://repository.iiitd.edu.in/xmlui/handle/123456789/388 2016-01-11T11:54:45Z