S Lab

[Sibin Mohan]

Systems Security Research Group at GWU and University of Illinois


  1. Essence/Cozart
  2. Kernel Fuzzing

Cozart/Essence

Team Members:

Collaborators:

Overview

Kernel debloating is a method of tracing memory objects to determine packages and libraries that are necessary to facilitate the usage of a custom application and building a custom kernel. This decreases the attack surface and increases the performance of these custom built applications in Linux environments as compared to running them in default Linux kernels.

Cozart is an automatic kernel debloating framework that addresses existing limitations by enabling dynamic kernel customization based on application needs. Experimental results demonstrate that Cozart significantly reduces kernel size while maintaining system stability and performance, outperforming traditional debloating techniques.

Essence, one of our current research projects, focuses on utilizing Cozart to build a more autonomous method to kernel debloating. Utilizing multiple frameworks for kernel tracing, including QEMU, and multiple forms of automation, it focuses on debloating the kernel up to 70-80% both statically and, more interestingly, dynamically via live tracing of the workload in order to minimize the necessary kernel size for execution. Future work in Essence focuses on running it in edge environments that may not have the resources that a full system may have, as well as running it in architectures beyond QEMU-emulated systems.


Kernel Fuzzing

Team Members:

Collaborators:

Overview

Past Team Members and Collaborators

Funding

This project is supported by funding from the Department of Energy (DoE)

(Prior) This project is was supported by a grant from the Office of Naval Reserach (ONR).

News