A public key infrastructure (PKI) provides secure communications between two different entities over an untrusted network.
Due to this ability, PKIs are now central to security on the Internet: there are a number of large-scale PKIs in use today such as DNSSEC, HTTPS, and the RPKI.
With multiple measurement techniques, I have taken a look at how PKIs are deployed, used, and often mismanaged: for example, I have analyzed large scale datasets such as all second level domains under TLDs or all SSL certificates served in IPv4, to understand PKI ecosystems on the Internet.
Recent years have seen the increased popularity and rapid growth in usage of multiple types of online social networks.
It is also becoming more easier to collect user's information from multiple sources and combining them to profile a specific person.
Exploiting abundant information obtained from online social network, we tried to attack and protect users from a such profiling attack.
For example, our study showed that a user's LGBT status can be accurately inferred with 93% accuracy, by applying a machine learning technique to predict users' personal traits such as sexual identities.
Over the past decade, online social networks have become major platforms to create and disseminate content, exchange opinion, share news, etc.
Due to their massive popularity, it is crucial to understand the contents that are shared by different users.
To better understand the characteristics of online content, we conducted measurement studies of two popular platforms: Pinterest and Reddit by measuring 700K threaded conversations from 1.5M users in Reddit and 1.6M pins shared by 1.1M users in Pinterest and provided valuable information for online retailers to enhance their marketing strategies.
Information Centric Networking (ICN) is one of the future Internet infrastructure that switches a host-centric paradigm (IP) to a content (data) paradigm.
Hence, it is crucial to build a commercial router that routes a traffic based on the content (or name).
We built a Named Data Networking (NDN) router, which is a scalable for routing and caching and proposed a design of an NDN router with unique forwarding and caching mechanisms featuring terabyte-scale caching with solid-state drives (SSD) while still forwarding packets at line speed and designed routing architecture to coordinate routing and caching in NDN to mitigate routing scalability and enhance the efficiency of the in-network storage.
P2P Networks solves many limitations derived from traditional server and client model.
BitTorrent, a popular peer-to-peer content distribution system, has received much attention due to its robustness and efficiency.
However, content has different popularity and availability over the network, hence, it is crucial to understand how contents are consumed to build more robust systems and distribute content more efficiently.
Due to its distributed nature, it is challenging to understand what kinds of content, when, and where they are shared without joining the peer-to-peer network.
To understand this, I developed a measurement framework monitoring more than 120K files shared by over 16M users in 168 countries.