huangyibo / bpf4ransomware

eBPF-based ransomware detection using machine learning

ebpfangel - Ransomware Detection using Machine Learning with eBPF for Linux

Authors: Max Willers, Tomás Philippart

Overview

This is the final result of an Offensive Technologies research project for the MSc Security and Network Engineering program of the University of Amsterdam. This is not intended to be a final product/solution to use in any productions environment whatsoever, it is simply the byproduct of research and therefore is intended to use as so.

Slides: https://github.com/TomasPhilippart/ebpfangel/blob/main/docs/ebpfangel-presentation.pdf

Paper: https://github.com/TomasPhilippart/ebpfangel/blob/main/docs/Ransomware_Detection_using_Machine_Learning_with_eBPF.pdf

Software architecture

flowchart LR
  subgraph sandbox_VM
    ransomware
  end
  ransomware --> network_hook & filesystem_hooks & crypto_hooks
  subgraph host_kernel
    network_hook & filesystem_hooks & crypto_hooks --> ebpf_c

  end
  subgraph host_userland
    ebpf_c -- events --> ebpf_python
    ebpf_python -- patterns + config --> ebpf_c
    ebpf_python --> machine_learning
  end

Loading

eBPF C program

flowchart LR
  A(event) --> B(compute stats <br/>per pid)
  B --> C(detect event pattern <br/>and threshold counts)
  X(python) -. patterns + config .-> C
  C --> D{report event?}
  D -- yes --> E(submit event <br/>to ring buffer) -->F(end)
  D -- no --> F

Loading

Learning resources and references

1. UNVEIL: A Large-Scale, Automated Approach to Detecting Ransomware
2. Toward A Network-Assisted Approach for Effective Ransomware Detection
3. Software-Defined Networking-based Crypto Ransomware Detection Using HTTP Traffic Characteristics
4. Ransomware Detection and Classification Strategies
5. Ransomware Detection techniques in the Dawn of Artificial Intelligence: A Survey
6. PayBreak: Defense Against Cryptographic Ransomware
7. Checking yourcryptography usage with eBPF (redhat, devconf 2020)
8. A Multi-Classifier Network-based Crypto Ransomware Detection System: A Casestudy of Locky Ransomware
9. A flow-based IDS using Machine Learning in eBPF
10. Kernel-level tracing for detecting stegomalware and covert channels in Linux environments
11. CryptoLock (and Drop It): Stopping Ransomware Attacks on User Data
12. https://www.bleepingcomputer.com/news/security/linux-version-of-rtm-locker-ransomware-targets-vmware-esxi-servers/ https://blogs.vmware.com/security/2022/02/avoslocker-modern-linux-ransomware-threats.html https://www.uptycs.com/blog/rtm-locker-ransomware-as-a-service-raas-linux (see yara rule)