Liwen Xu (许力文)

Research Projects

National Student Innovation Funding Joint Research Program (Sep 2022 - Oct 2023)

• Name: Research on Localized Intelligence Production and Threat Discovery Based on Network Endogenous Data to Discover Unknown Attacks
• Consigner: QI-ANXIN Technology Group Inc.
• Role: Project Leader
• Dynamic Graph Modeling: Innovated a dynamic graph model for complex networks using multi-source data, significantly enhancing real-time attack detection and reducing computational overhead.
• Hierarchical Neural Network for Threat Correlation and Discovery: Created HiSec, a cutting-edge hierarchical neural network framework, improving the detection of advanced cyber threats and enriching threat intelligence.
• Attack Graph-Based Threat Intelligence: Formulated the Ag2Intel model to transform attack graphs into actionable threat intelligence, elevating the accuracy and readability of security insights.

China Mobile Shanghai Company Carrier-Class Data Security System Evaluation Research Project (Confidential, Jul 2022 - Jun 2023)

• Name: Analysis and Research on Data Security Situational Awareness and Disposal Mechanism
• Consigner: China Mobile Shanghai Company
• Role: Joint Researcher

National Natural Science Foundation of China Joint Key Project (Confidential, Sep 2021 - Present)

• Name: Intelligent Detection and Defense of Highly Concealed Unknown Attacks on Critical Infrastructure
• Advisor: Prof. Jianhua Li
• Role: Researcher of Correlation Analysis for Unknown Attacks

AI & Cybersecurity Innovation Platform of Shanghai Jiao Tong University (Sep 2021 - Present)

• Goal: Construction of FinTech Testing Range and Blockchain Security Sub-Platform of Critical Information Infrastructure
• Role: Core Member
• Blockchain Security Range Construction: Actively supported the creation of a blockchain security testing environment, facilitating the rapid deployment of private blockchain networks. Enhanced the platform's capabilities with critical smart contract vulnerability detection, reproduction, and remediation tools. Developed simulation tools for Proof of Work (PoW) consensus protocol attacks and mining behavior detection and analysis.
• FinTech Security Range Development: Engaged in the construction of a FinTech testing range aimed at researching foundational scientific issues such as cyber mimicry simulation for critical infrastructure networks and tracking of network attack behaviors. The range is designed to support key research activities in the area of network security and attack traceability within the financial technology sector.

Transverse Research Project (Confidential, Sep 2021 - Jan 2023)

• Name: Research and Application of Cloud Network Security Situational Awareness and Threat Warning
• Consigner: State Grid Shanghai Municipal Electric Power Company, Shanghai
• Role: Key Researcher of Cloud Security Threat Analysis

Undergraduate Graduation Project (Thesis) (Sep 2020 - May 2021)

• Name: A Real-time Intelligent Detection System for Malicious Network Traffic Based on Deep Learning
• Award: Outstanding Thesis Award of Xidian University
• Thesis Overview: Developed an innovative real-time malicious traffic detection system for my bachelor's thesis, leveraging the Storm big data streaming framework combined with deep learning technology.
• Achievement: Developed an efficient real-time preprocessing algorithm compatible with the Storm streaming framework to transform network traffic into the required data structure for the model. The deep learning model demonstrated a high overall accuracy of 97.36% for a 20-class classification task on a small dataset, with individual class metrics exceeding 90% and the majority surpassing 98%.
• Impact: The integration of big data architecture and deep learning algorithms resulted in a system that significantly outperforms traditional methods in detection performance and real-time efficiency

The 5th Chinese National Cryptography Technology Competition (May 2019 - Nov 2019)

• Name: RISC-V Processor Core for Chinese National Cryptographic Algorithm
• Award: Special Prize at the 5th Chinese National Cryptography Technology Competition, China
• Role: Team Leader
• Project Overview: Achieved top honors for designing a RISC-V processor tailored to Chinese National Cryptographic Standards, specifically focusing on the structural features of the SM3 and SM4 algorithms.
• Research and Development: Conducted in-depth research into the SM3 and SM4 algorithms to identify critical areas for acceleration. Designed custom acceleration instructions within the RISC-V instruction set architecture.
• Impact: The enhanced Xcore processor facilitated accelerated cryptographic instruction execution and improved instruction speedup ratio and algorithm throughput without additional hardware investment.