PUL-Inter-slice Defender: An Anomaly Detection Solution for Distributed Slice Mobility Attacks

TL;DR

This work tackles Distributed Slice Mobility attacks that exploit inter-slice switching in 5G network slicing by introducing PUL-Inter-Slice Defender, a positive-unlabeled learning framework. It couples an LSTM-Autoencoder for temporal latent feature extraction with K-Means clustering, guided by 3GPP KPI and PM-counter data to detect RSA and TSA under contaminated training data. Evaluated on a real 5G testbed (free5GC+UERANSIM), the method achieves F1-scores above 98% across contamination levels and outperforms the original Inter-Slice Defender and two PUL baselines, demonstrating robustness to data contamination. The paper also discusses deployment considerations with NWDAF and proposes privacy-preserving extensions like federated learning, outlining future work on concept drift and scaled NS environments.

Abstract

Network Slices (NSs) are virtual networks operating over a shared physical infrastructure, each designed to meet specific application requirements while maintaining consistent Quality of Service (QoS). In Fifth Generation (5G) networks, User Equipment (UE) can connect to and seamlessly switch between multiple NSs to access diverse services. However, this flexibility, known as Inter-Slice Switching (ISS), introduces a potential vulnerability that can be exploited to launch Distributed Slice Mobility (DSM) attacks, a form of Distributed Denial of Service (DDoS) attack. To secure 5G networks and their NSs against DSM attacks, we present in this work, PUL-Inter-Slice Defender; an anomaly detection solution that leverages Positive Unlabeled Learning (PUL) and incorporates a combination of Long Short-Term Memory Autoencoders and K-Means clustering. PUL-Inter-Slice Defender leverages the Third Generation Partnership Project (3GPP) key performance indicators and performance measurement counters as features for its machine learning models to detect DSM attack variants while maintaining robustness in the presence of contaminated training data. When evaluated on data collected from our 5G testbed based on the open-source free5GC and UERANSIM, a UE/ Radio Access Network (RAN) simulator; PUL-Inter-Slice Defender achieved F1-scores exceeding 98.50% on training datasets with 10% to 40% attack contamination, consistently outperforming its counterpart Inter-Slice Defender and other PUL based solutions combining One-Class Support Vector Machine (OCSVM) with Random Forest and XGBoost.

Figures (7)

• Figure 1: Overview of DSM attack variations, illustrating how compromised UEs initiate ISS to perform RSA and TSA, resulting in excessive signaling load on 5G Core NFs. The colored line indicates the targeted NSs.
• Figure 2: Architecture of Inter-Slice Defender and PUL-Inter-Slice Defender. Inter-Slice Defender uses an LSTM-Autoencoder trained on benign data to detect attacks via reconstruction error and a threshold, while PUL-Inter-Slice Defender adopts PUL by leveraging the latent space for feature extraction to classify benign and DSM attack samples in the presence of contaminated training data through K-means.
• Figure 3: 5G testbed comprising four NSs, designed in accordance with the 3GPP standard.
• Figure 4: 2D PCA plot showing positive and negative classes in the training dataset with different levels of negative samples
• Figure 5: AMF CPU consumption during benign and attack emulations.