LOHIT: An Online Detection & Control System for Cellular SMS Spam

S. Dixit, S. Gupta, and C.V. Ravishankar (USA)


Spam, short text message (SMS), SMSC, mobile, Clustering


The efficient and accurate control of spams on mobile handsets is an important problem. Mobile spam incurs a cost on a per-message basis, degrades normal cellular service, and is a nuisance and breach of privacy. It is also a popular enabler of mobile fraud. In countries such as South Korea and Japan, Mobile Spamming generates almost half of the total SMS traffic. In this paper we propose a novel spam control technique based on random projections, designed to run on SS7 links so that spams are supressed before they reach users. Our is a non Bayesian, non–keyword approach, which rate limits candidate spam messages to foil spammers. We demonstrate using real-world spam messages that random projection is a robust, efficient and accurate method to identify SMS spam. We give a mathematical formulation of the SMS spam problem and demonstrate that it models the real world short message spam paradigm accurately. Based on this formulation, we describe a framework and algorithm to efficiently identify and control spam messages at the SMSC switch of a mobile network.

Important Links:

Go Back