In this paper, we propose a machine learning methodology for prediction of signaling sessions established with the Session Initiation Protocol (SIP). Given the increasing importance of predicting and detecting abnormal sequences of SIP messages to avoid SIP signaling-based attacks, we first propose a Bayesian inference method capable of representing the statistical relation between a SIP message, observed by a SIP user agent or a SIP server, and prior trustworthy SIP dialogs. The Bayesian inference method, a Hidden Markov Model (HMM) enriched with n−gram Markov observations, is updated over time, so the inference can be used in real-time. The HMM is then used for predicting and detecting SIP dialogs through a lightweight implementation of Viterbi algorithm for sparse state spaces. Experimental results are also reported, where a SIP dataset representing prior information collected by a SIP user agent and/or a SIP server is used to predict or detect if a received sequence of SIP messages is legitimate according to similar SIP dialogs already observed. Finally, we discuss the results obtained for a dataset of abnormal SIP sequences, not observed during the inference stage, showing the effective utility of the proposed methodology to detect abnormal SIP sequences in a short period of time.

No abstract

This paper investigates the detection of abnormal sequences of signaling packets purposely generated to perpetuate signaling-based attacks in computer networks. The problem is studied for the Session Initiation Protocol (SIP) using a dataset of signaling packets exchanged by multiple end-users. The paper starts to briefly characterize the adopted dataset and introduces a few definitions to propose a deep learning-based approach to detect possible attacks. The solution is based on the definition of an orthogonal space capable of representing the sampling space for each time step, which is then used to train a recurrent neural network to classify the type of SIP dialog for the sequence of packets observed so far. When a sequence of observed SIP messages is unknown, this represents possible exploitation of a vulnerability and in that case, it should be classified accordingly. The proposed classifier is based on supervised learning of two different sets of anomalous and non-anomalous sequences, which is then tested to identify the detection performance of unknown SIP sequences. Experimental results are presented to assess the proposed solution, which validates the proposed approach to rapidly detect signaling-based attacks.

This paper investigates the detection of abnormal sequences of signaling packets purposely generated to perpetuate signaling-based attacks in computer networks. The problem is studied for the Session Initiation Protocol (SIP) using a dataset of signaling packets exchanged by multiple end-users. A sequence of SIP messages never observed before can indicate possible exploitation of a vulnerability and its detection or prediction is of high importance to avoid security attacks due to unknown abnormal SIP dialogs. The paper starts to briefly characterize the adopted dataset and introduces multiple definitions to detail how the deep learning-based approach is adopted to detect possible attacks. The proposed solution is based on a convolutional neural network capable of exploring the definition of an orthogonal space representing the SIP dialogs. The space is then used to train the neural network model to classify the type of SIP dialog according to a sequence of SIP packets prior observed. The classifier of unknown SIP dialogs relies on the statistical properties of the supervised learning of known SIP dialogs. Experimental results are presented to assess the solution in terms of SIP dialogs prediction, unknown SIP dialogs detection, and computational performance, demonstrating the usefulness of the proposed methodology to rapidly detect signaling-based attacks.