Biography
Dr. Kesheng (John) Wu leads multiple R&D endeavors focused on advanced technologies and testbeds at the Scientific Networking Division of Lawrence Berkeley National Laboratory. These projects aim to expedite data transfer among DOE user facilities, implement in-network storage and computational resources for intricate scientific workflows, and explore algorithms, strategies, and practices to enhance the efficiency of network operations. Additionally, Dr. Wu's team is tasked with developing and managing networking testbeds, providing the broader research community with platforms to explore future generations of networking technologies and optimize their utilization. These testbeds encompass conventional optical networking alongside cutting-edge quantum communication capabilities.
SRP Project Title
Advanced Modeling for Prediction and Anomaly Detection in Network Operations
Topical Areas
Applied Computer Science; Artificial Intelligence and Intelligent Systems; Infrastructure and Instrumentation
Abstract
As we approach the summer of 2026, our goal is to develop and deploy advanced modeling capabilities for predicting hardware failures and detecting external configuration anomalies in our network infrastructure. This project aims to leverage machine learning and data analytics to improve network stability, reduce unplanned outages, and enhance overall network efficiency. We will focus on two key areas: (1) Predicting Hardware Failures: By analyzing log and telemetry data, we will develop models to predict hardware failures in networking equipment, enabling proactive replacement and minimizing downtime. (2) Detecting Network Configuration Anomalies: We will create a tool to analyze received network configuration data, identify routing path changes, and notify engineers of anomalies, ensuring swift response to potential disruptions. Through this project, we will apply advanced modeling techniques to drive business outcomes, including: - Improved network stability and reduced unplanned outages - Enhanced predictive capabilities for hardware failures and configuration anomalies - Data-driven decision-making for network operations and maintenance By investing in advanced modeling and analytics, we will take a proactive approach to network operations, ensuring a more resilient and efficient network infrastructure for the summer of 2026 and beyond.
Desired Skills
Strong analytics skills are must. Computer networking knowledge is desired.
Lightning Talk Title
Advanced Modeling for Prediction and Anomaly Detection in Network Operations
Keywords
Predictive Modeling Anomaly Detection Network Stability Machine Learning Data Analytics Network Operations