Albert Kwon, a Ph.D. student at the Massachusetts Institute of Technology (MIT) has been thinking about ways to develop more robust security methods for anonymous networks since late 2013. He and David Lazar, another graduate student at the university, worked together on an applied cryptography project as an assignment for a graduate course, and their conversations ultimately grew into an idea for a cryptographic method that may extend the anonymity promises of the Tor network.
Their conversations grew into a research paper titled “Riffle: An Efficient Communication System With Strong Anonymity,” that demonstrates a new methodology of sending information through an anonymous network. The anonymity network methodology, dubbed Riffle, expects to be more efficient and secure than existing approaches.
The research demonstrates a proof-of-concept for a methodology that Kwon said would protect anonymity even against a state-level adversary. An adversary capable of monitoring network traffic could ascertain which servers are communicating with each other by using traffic analysis. Tor is weak at protecting against this method, said Kwon, lead author of the report.
The researchers published the report soon after the legal case heated up around the FBI's use of a “network investigative technique” to identify the IP addresses of visitors to the child pornography forum Playpen.
While the method does provide strong protections, Kwon was careful to note that Riffle may not protect against some of the exploits that were used by the FBI in the Playpen case, such as browser-level malware.
Kwon and Lazar began working together after the Snowden disclosures. Kwon said the leaks certainly raised his interest in cryptology. “We are trying to provide network level protection, not necessarily JavaScript sanitization,” he told SC Media.
Co-authors of the research paper beside Kwon and Lazar include Srinivas Devadas, Kwon's adviser; and Bryan Ford, an MIT alumni and professor at the École Polytechnique Fédérale de Lausanne.
Whereas Tor originally provided no protection against malicious servers, the concept behind Riffle assumes servers could be malicious. The approach makes use of a series of servers, or a mixnet, and rearranges the order in which messages are sent to each server. “Even if there are adversarial servers, if there is one honest server, the security will be protected,” says Kwon. “We wanted to make it as usable as possible.”
While the proof-of-concept for the anonymity methodology is not yet available as a product, it has been frequently compared to the Tor protocol. Kwon sees the Riffle method as a “complementary” approach to the anonymity protocol. While Tor is designed for everyday use, an individual would use Riffle “if the government has been tracking you and you really want to stay secure.”
He noted that he is still working to solve protocol-level challenges. The prototype does not handle clients leaving and entering the network at any time very well.
Kwon tells SC Media that developing the Riffle method into an operational protocol would require more engineering efforts by software engineers. He is considering releasing an “alpha-level” prototype and hopes to release the initial version next semester.Please click hear to read about the next Thinker: Titus Melnyk, head of IT security operations and technology, FCA U.S.
