Leveraging Email based Social Networks to Prevent Spam: Framework, System Design and Evaluation

Abstract

Spam is still an open problem from the network operator's perspective. The common state-of-the-art strategy is to place filters against spam at the recipient's edge. Although this strategy largely solves the spam problem from the user's perspective but the false positives/negatives may still exist. This strategy is also unable to prevent spam from traversing the Internet. Consequently, we now have around 260 billion spam messages sent across the Internet each day. Spam is therefore consuming large amounts of Internet bandwidth and is imposing non-negligible financial loss to network operators. Therefore it becomes imperative to mitigate spam much earlier than is typically done today. The main part of this thesis proposes LENS, a novel, easily adaptable and scalable spam protection system that is incrementally deployable with low processing overheads. LENS leverages the recipient's social network to allow correspondence within the social network to directly pass to the mailbox of the recipient. LENS further mitigates spam beyond social circles and stops spam messages early on, instead of filtering these messages from user mailboxes or at the recipient's edge. The key idea in LENS is to select legitimate and authentic users, called Gatekeepers (GKs), from outside the recipient's social circle and within predefined social distances. LENS utilizes the GK to generate a voucher, and new senders are required to obtain these vouchers (to communicate with the recipient) from a GK in their social neighborhood. Recipients recover from compromised GKs simply by selecting replacements and revoking vouchers. Unless a GK vouches for the emails of potential senders from outside the social circle of a particular recipient, those e-mails are prevented from transmission. In this way LENS drastically reduces the consumption of Internet bandwidth by spam. The contributions of this thesis are the development of social network based spam mitigation framework (LENS), a system design based on this framework, the evaluation of the system by means of simulation and a prototype implementation. The evaluations show that with the help of hundreds of GKs, LENS can provide reliable email delivery from millions of potential users. LENS imposes Zero overhead for the common case of frequent and familiar senders, and remains lightweight for the general case. Using real email traces, the simulations show that LENS is effective in accepting all legitimate inbound emails. Our prototype implementation of LENS in Postfix/MailAvenger shows that LENS consumes up to 75% less CPU, 9% less memory and it is around 2-3 orders of magnitude faster in processing emails than traditional solutions like SpamAssassin.