ISRL

• Funding source: National Research Foundation
• Program: Mid-Career Researcher Program Type 1-2
• Period: 2023.3.1-2028.2.29
• Budget: One billion KRW for five years

Research statement: From a network security perspective, one of the most important recent trends is “In-Network Defense”. In addition to endpoint security and perimeter security, which have been the focus to date, there is a new trend of enabling security functions in the backbone network. Moving away from the current perspective that defines the role of the network as performing only routing functions, the network is being asked to perform more advanced functions from a security perspective, which essentially foreshadows the emergence of routers equipped with artificial intelligence models. Along with the rapid development of deep learning in most fields of IT and the growing importance of data analysis using AI/ML technologies, various attempts have been made to enable routers with AI models to analyze network traffic on a per-flow basis to perform security functions. However, performing AI models within the router without delaying the delivery of large numbers of high-speed packets is not feasible by current technologies. Therefore, it is our goal to develop a self-driving router that performs AI models in real time and autonomously performs security-related functions.

Me, Jiyoo, Jinchun at a restaurant in UCF, Orlando, FL, Feb. 7th, 2018 (I visited UCF for GRL workshop)

I am delighted to share this good news. Jinchun is going to join ETRI (Electronics and Telecommunications Research Institute) in January 2023. Big congratulations to Jinchun!

He was one of my PhD students at Inha and also was co-advised by David@UCF starting in 2018. He obtained PhD degree (CS) in 2020, and then he worked as a post-doctoral researcher in Texas A&M University-Kingsville, TX, USA ( 2020 – 2022 ). His main research theme is block chain, IoT security, malware analysis, and Cyber-Physical System security. He is a hard-working researcher and successfully published 21 papers in top journals and conferences including Computer Networks, IEEE transactions on network and service management, IEEE IoT journal, IEEE Systems Journal, etc. Now, he is going to start another stage in his career in DaeJeon. I believe that he is gonna be good and successful as a researcher.

Congrats again Jinchun!

Sian gave a talk on PortCatcher in CCS 2022 in LA, US on the 10th of November, 2022.

A photo of the room “Museum A”. Sian is on the podium, and I was taking a photo of her (Left).

Rhongho, David, and me at a hotel lobby (Photographer was JeongHye.) It’s been quite a long time to gather together.

Sian, Juhee, JeongHye, and me in the banquet

Angels flight railway was just by our hotel. A scene from LaLa Land was filmed here.

이화여대 정보보호연구실에서는 학부생 인턴, 석사, 박사, 석박사 통합과정 학생들을 모집하고 있습니다. AI/딥러닝 기반 보안, 네트워크/시스템 보안, 블록체인 등에 관심있는 학생들은 nyang@ewha.ac.kr로 연락바랍니다. (혹은 연구실 랩장 김시안 ksy60a@gmail.com 에게 연락 주세요)

We always welcome students who are interested in research on AI/Deep-Learning-based security and network/system security. If you are interested in joining our lab to continue study, don’t hesitate to contact me at nyang@ewha.ac.kr. (or you can contact Laboratory Manager: Sian Kim, ksy60a@gmail.com)

Another big news. Our paper titled “A Robust Counting Sketch for Data Plane Intrusion Detection”, proposing Count-Less, is accepted in NDSS 2023! This paper is part of the series of works that our group has been pursuing on data sketch for traffic measurement and security applications. Count-Less is a major variation of the famous Count-Min sketch, but is more robust in terms of measurement accuracy under a various traffic distributions, including attack traffic. The design of the sketch is lightweight enough to fit in the packet processing pipeline and sketch’s performance is shown to be superior to state-of-the-art algorithms, such as Elastic sketch and FCM by comprehensive experiments.

Congratulations to Sian, Changhun, Rhongho, David, and me!

Kichang Yang at Hyundai motor company gave a talk on vehicular security in cyber security department’s weekly colloquium at Ewha. His talk showed many interesting vehicular security issues as well as what to do to become a vehicular security engineer. It must be very helpful to young students as a career guidance. He has been leading a security group as a director of security division in HMC since 2016. His group is working on all aspects of vehicular security like hacking, reproducing reported vulnerabilities, regulations, and what not. We got to understand what are the issues in vehicular security, career path, and even how to do a good interview.

Every week, we will have an invited talk about various issues in security!

I am pleased to announce that Dr. Changhun Jung starts his career as a researcher at R&D center of Hyundai Motor Company.

He obtained M.S. and PhD degree in 2017 and 2022, respectively in my lab. His main research interests are authentication, network security, data plane security, AI-based attack detection, data structure and algorithm for high performance switch architecture. During his study, he published eight papers in top venues such as ICASSP (2018), AAAI AI for Social Good (2019), Computers and Security (2021), BMC Medical Imaging (2022), IEEE/IFIP DSN (2022), IEEE TMC (to appear), ACM CCS (2022), and NDSS (2023). He is one of the most earnest students in pursuing research goals I have ever seen, and I was able to get through a lot of difficulties we met during our research work owing to his earnest endeavors. Even though he is leaving us, we are going to collaborate on further research works.

Congratulations again to Changhun!

I am pleased to share that our paper “A Scalable and Dynamic ACL System for In-Network Defense” (named PortCatcher) is accepted in CCS 2022! A big achievement of our research group.

Summary by David: This work addresses the scalability of dynamic ACL systems through an elegant combination of algorithmic optimization and new architectural design that exploits SRAM instead of TCAM for ACL storage. The work is implemented in P4 and this coupling allows us to process packets at the line speed, even for orders of magnitude growth in size of the ACL in contrast to TCAM-based approach.

Congratulations to Changhun Jung, Sian Kim, Rhongho Jang, David Mohaisen, and Me.

More news will come soon. 😉

I am pleased to announce that our paper “Systematically Evaluating the Robustness of ML-based IoT Malware Detection Systems” authored by A. Abusnaina, A. Anwar, S. Alshamrani, A. Alabduljabbar, R. Jang, D. Nyang, D. Mohaisen is accepted (just passed the shepherding) in the 25th International Symposium on Research in Attacks, Intrusions and Defenses (RAID 2022).

This paper deals with the issue on the performance of ML-based IoT malware detectors against malware mutations. Results show that even the state-of-the-art technologies are suffering from incompetence in detecting even simple mutations like functionality-preserving operations (packing, stripping, and padding). This calls for research community to put their efforts towards testing the robustness of malware detectors.

Congratulations to all!

Our paper titled “WBC image classification and generative models based on convolutional neural network” is published in BMC Medical Imaging journal.This paper deals with the white blood cell classification and the synthesis of the images. Our AI model based on the simple CNN, called W-Net outperforms the famous AlexNet, VGGNet, and ResNet50/18 by far in terms of accuracy and computation complexity. We explore why this simple architecture works better than those state-of-the-art technologies for WBC classification. Also, it is well-known to be quite difficult to obtain high quality WBC images with labels even for research purporses, and thus, we release the high quality synthesized WBC images generated from real hospital WBC images to aid and expedite research in this area. Check out our paper.

Congratulations to ChangHun, Mohammed, David, KyungJa, and me!