Malicious SIM Cards Identification Method Based on Telecom Big Data Analytics

Xiaoling Feng, Tianming Zhang, Yue Zhao, Zhi Yu, Long Yan, Bo Chen, Yulun Song

Abstract: This paper presents a method that helps e-commercial enterprises enhance risk control capabilities and operate anti-fraud during online marketing events. By definition, Malicious SIM Cards (MSCs) identification in this paper aims to identify SIM cards which are utilized by malicious purposes on specified terminals at scale (e.g., SIM modem pool or bulk SMS machine), and behave significantly different from normal SIM cards, especially towards e-commercial companies' marketing events, e.g., releases of coupons. The proposed MSC identification method is implemented as a label-function-based factor matrix framework. Particularly, a set of label functions is constructed from six dimensions via telecom big data analytics, including frequency of replacement, location, card type, Call Detail Records (CDRs), ID type, in-network-period in terms of SIM cards. Subsequently, MSC labels are provided to the identified SIM cards among the entire China Unicom SIM cards, which realize an effective way for large scale commercial use. Finally, selected features' distribution of MSCs identified and normal SIM cards are illustrated for comparison purposes.