A5051468807- Network Traffic and Congestion Control
- Software System Performance and Reliability
- Traffic Prediction and Management Techniques
- Software-Defined Networks and 5G
- Network Security and Intrusion Detection
- Image and Video Quality Assessment
- Cloud Computing and Resource Management
- Data Visualization and Analytics
- Internet Traffic Analysis and Secure E-voting
- Vehicle License Plate Recognition
- Caching and Content Delivery
- Wireless Networks and Protocols
- Robotic Path Planning Algorithms
- Time Series Analysis and Forecasting
- Network Packet Processing and Optimization
University of Illinois Urbana-Champaign
2021-2024
Amrita Vishwa Vidyapeetham
2022
Microsoft Research (India)
2020
Bandwidth estimation and congestion control for real-time communications (i.e., audio video conferencing) remains a difficult problem, despite many years of research. Achieving high quality experience (QoE) end users requires continual updates due to changing network architectures technologies. In this paper, we apply reinforcement learning the first time problem (RTC), where seek optimize user-perceived quality. We present initial proof-of-concept results, learn an agent sending rate in RTC...
As modern cloud services have scaled out, applications moved from relatively monolithic designs to highly modularized fleets of microservices that communicate among each other perform application-level tasks. These effectively form a network at the application layer, and service mesh frameworks recently emerged factor out microservices' common communication functionality.
Modern cloud-based applications have complex inter-dependencies on both distributed application components as well network infrastructure, making it difficult to reason about their performance. As a result, rich body of work seeks automate performance diagnosis enterprise networks and such cloud applications. However, existing methods either ignore which results in poor accuracy, or require causal acyclic dependencies cannot model common environments.
We present a vision of data-informed network simulation to address significant shortcomings in the state art. substantiate our position with proof points based on iBox, which leverages networking domain knowledge and machine learning (ML) models, coupled plentiful data, provide pathway perpetual renewal simulators.
While network simulation is widely used for evaluating protocols and applications, ensuring realism remains a key challenge. There has been much work on simulating mechanisms faithfully (e.g., links, buffers, etc.), but less attention the critical task of configuring simulator to reflect reality. We present iBox ("Internet in Box"), which enables data-driven path simulation, using input/output packet traces gathered at sender/receiver target create model end-to-end behaviour path. Our builds...
While network simulation is widely used for evaluating protocols and applications, ensuring realism remains a key challenge. There has been much work on simulating mechanisms faithfully (e.g., links, buffers, etc.), but less attention the critical task of configuring simulator to reflect reality.
While network simulation is widely used for evaluating protocols and applications, ensuring realism remains a key challenge. There has been much work on simulating mechanisms faithfully (e.g., links, buffers, etc.), but less attention the critical task of configuring simulator to reflect reality. We present iBox ("Internet in Box"), which enables data-driven path simulation, using input/output packet traces gathered at sender/receiver target create model end-to-end behaviour path. Our builds...
With the rapidly growing number of hosts connected to internet, there is an ever-increasing demand for fast and inexpensive switch memory. At same time, network functions handled at switch, especially in case a programmable increasing steadily (e.g., purposes routing, telemetry, load balancing), which require dedicated Various compact efficient data structures Bloom filters [15], ludo hashes [10], cuckoo [3]) have been proposed past address these needs. However, can provide varying...