I am currently a Research Scientist at Meta working on AI Privacy. I received my PhD degree from the department of Computer Science, Stony Brook University, New York, where I worked on the Network Security and Applied Cryptography (NSAC) lab under the guidance of Prof. Radu Sion.
My area of interests are system security, data privacy, embedded storage system, machine learning and clouds.
Before coming to US, I received my Bachelor degree at Tsinghua University, China.
Doctor of Philosophy (Ph.D.), Computer Science Stony Brook University
Advisor: Radu Sion
Bachelor's Degree, Automation Tsinghua University
Advisor: Jie Zhou
SOK: Plausibly deniable storage
Chen Chen, Xiao Liang, Bogdan Carbunar, Radu Sion [pdf] [bibtex]
Proceedings on Privacy Enhancing Technologies 2022 (PETS 2022)
PEARL: Plausibly Deniable Flash Translation Layer using WOM coding
Chen Chen, Anrin Chakraborti, Radu Sion [pdf] [bibtex]
USENIX Security Symposium 2021 (USENIX Security 2021)
INFUSE: Invisible plausibly-deniable file system for NAND flash
Chen Chen, Anrin Chakraborti, Radu Sion [pdf] [bibtex]
Proceedings on Privacy Enhancing Technologies 2020 (PETS 2020)
PD-DM: An efficient locality-preserving block device mapper with plausible deniability
Chen Chen, Anrin Chakraborti, Radu Sion [pdf] [bibtex]
Proceedings on Privacy Enhancing Technologies 2019 (PETS 2019)
DataLair: An Efficient Block Device Mapper with Plausible Deniability
Anrin Chakraborti, Chen Chen, Radu Sion [pdf] [bibtex]
Proceedings on Privacy Enhancing Technologies 2017 (PETS 2017)
KXRay -- Introspecting the kernel for rootkit timing footprints
Chen Chen, Darius Suciu, Radu Sion [pdf] [bibtex]
23rd ACM Conference on Computer and Communications Security (CCS 2016 Poster)
DataLair - A Storage Block Device with Plausible Deniability
Anrin Chakraborti, Chen Chen, Radu Sion [pdf] [bibtex]
23rd ACM Conference on Computer and Communications Security (CCS 2016 Poster)
Quantitative Musings on the Feasibility of Smartphone Clouds
Chen Chen, Moussa Ehsan, Radu Sion [pdf] [bibtex]
15th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing (CCGrid 2015)
HIFS: History Independence for File Systems
Sumeet Bajaj, Chen Chen, Abhishek Kumar, Radu Sion [pdf] [bibtex]
24th ACM Symposium on Operating Systems Principles (SOSP 2013 Poster)
PEARL is a plausibly deniable FTL for NAND flash memory. It allows both public data and hidden data to coexist in one flash device without leaking the existence of the hidden data to multi-snapshot adversaries.
INFUSE is a plausible deniability system for raw NAND flash devices. It hides sensitive data in flash devices by “embedding” logical data using variations in threshold voltage. The hidden data stored in the device and the deployment of INFUSE can be both “invisible” to adversaries.
Datalair is a practical plausible deniability system built based on a new write-only ORAM. It reduces the complexity of the state of the art existing write-only ORAM by a factor of O(logN). When compared with existing approaches, DataLair is two orders of magnitude faster (and as efficient as the underlying raw storage) for public data accesses, and 2-5 times faster for hidden data accesses.
PD-DM is a new efficient device mapper with strong plausible deniability against multi-snapshot adversaries. It preserves locality and increases performance by ensuring most of its underlying writes are sequential. In a typical setup, throughputs are one order of magnitude (10-100x) faster than existing approaches.
This project aims to detect the existence and location of specific instances of target data structure types in a VM by observing memory accesses and training for targetspecific timing-based signatures. We deploy the detection mechanisms to defeat kernel rootkits that "hide" their associated processes from existing snapshot-based detection methods. We introduce multiple signature variants and evaluate them for different kernel versions.
Smartphone Datacenter looks insight into the power-performance tradeoff at scale for ARM and x86 architectures by quantifying the cost/performance ratio precisely enough to allow for a broader conclusion about the feasibility of deploying an ARM datacenter in next few years.
Green DIMM is a system that aims at energy efficient memory management in OS.
HIFS for flash is a file system designed for flash storage devices with a good balance between the history independent security and device life time.