Authors:
Gabriel Jozsef Berecz
1
and
Istvan-Gergely Czibula
2
Affiliations:
1
Department of Computer Science, Babes¸-Bolyai University, M. Kog˘alniceanu Street, Cluj-Napoca, Romania, Cyber Threat Proactive Defense Lab, Bitdefender and Romania
;
2
Department of Computer Science, Babes¸-Bolyai University, M. Kog˘alniceanu Street, Cluj-Napoca and Romania
Keyword(s):
Cryptojacker, Computer Security, Learning-based Classifier.
Related
Ontology
Subjects/Areas/Topics:
Information and Systems Security
;
Intrusion Detection & Prevention
Abstract:
Cryptocurrencies are renowned world wide nowadays and they have been adopted in various industries. This
great success comes from both the technology innovation they brought to the world, the blockchain, and the
financial opportunities they opened up for investors. One of the unpleasant aspects are the cybercriminals who
took advantage of this technology and have developed malicious software (i.e. cryptojacker) in order to gain
profit by mining cryptocurrencies on their victims’ personal computer without any consent. This paper proposes
to analyze standalone cryptojackers, both statically and dynamically, with the aim of identifying specific
traits. The approach draws out features specific to cryptojackers that are selected using statistical methods and
explains why a cryptocurrency mining malware has such traits. Based on 20 selected specific features, three
different supervised learning classification models have been trained, which are able to differentiate between
clean
applications and cryptojackers reliably. In experiments, an average accuracy of 92.46% has been
achieved.
(More)