CNI CONNECT

Monthly Newsletter from CNI

Issue - March 2025

Upcoming Events

We are excited to launch Network In Code (NIC), a two-week coding competition in Networking & Systems, organized by CNI, IISc, co-hosted by COSH, NITK, and supported by Cisco CSR.

Participants will tackle real-world challenges in eBPF, network simulations, containers, Kubernetes, network security, and IPv6, gaining hands-on experience.

🏆 Prizes worth INR 2,00,000 and an internship at IISc await the top performers!

📌 Register now and stay updated on our website!

What’s cooking at CNI?

Expanding Horizons: Research Talks and Discussions at CNI

In February, CNI had the privilege of hosting Prof. Sid Jaggi (University of Bristol), Prof. Koushik Kar (Rensselaer Polytechnic Institute), and Dr. Divyanshu Pandey (Rice University). During their visit, they engaged with IISc students, participated in research discussions, and explored future directions in their respective fields. As part of their visit, they also delivered talks on their ongoing research: Prof. Sid Jaggi presented on Group Testing, Prof. Koushik Kar discussed Traffic Peering Games in Internet Exchange Points and Dr. Divyanshu Pandey spoke on Exploring multi-bounce scattering to increase the field-of-view of mm-wave radar.

Advances in Group Testing

Prof. Jaggi discussed group testing, a method for identifying a small subset of individuals or items of interest within a large population using pooled tests. This technique has applications in medical diagnostics, RFID systems, industrial quality control, and streaming algorithms. He provided an overview of fundamental bounds and algorithms for various models and presented recent advancements in the field, highlighting its relevance to broader non-linear sparse inverse problems.

Traffic Peering Games in Internet Exchange Points

Prof. Kar discussed the dynamics of traffic exchange at Internet Exchange Points (IXPs), where ISPs establish peering connections. Using game-theoretic models, he analyzed how pricing strategies and capacity decisions impact efficiency, showing that congestion-proportional pricing maintains near-optimal performance. He also explored machine learning-based automation of peering decisions and highlighted broader applications of these models in network traffic, power, and heat flow management.

Exploring Multi Bounce Scattering to increase the field-of-view of mm wave radar

Dr. Pandey presented a novel approach for mm-wave radars to detect objects beyond their direct field of view by leveraging multipath signal reflections. Unlike traditional radar processing, which treats such reflections as noise, this method explicitly models them, expanding radar perception without additional hardware or prior environment knowledge. The technique, tested on a commercial radar platform, has potential applications in autonomous navigation, disaster management, and joint communication-sensing systems.

Group Testing: Something old, something new, something borrowed

Prof. Nikhil Karamchandani delivered an insightful talk on Group Testing, a technique for efficiently identifying defective items within a large population using minimal pooled tests. While originally developed for medical diagnostics, group testing has broader applications in wireless communications, DNA sequencing, network tomography, and neighbor discovery.

The talk covered key theoretical results and introduced Cascaded Group Testing, a new variation where tests return the first defective item in an ordered subset. This approach offers exciting possibilities for optimizing testing strategies across various fields.

Data-Driven Next-Generation Wireless Networking: Harnessing AI for Superior Performance and Security

Prof. Prasenjit Chanak discussed the role of Artificial Intelligence (AI) in future wireless networks, addressing challenges in IoT, 5G, and NextG systems, including high data rates, scalability, security, and cost-effective deployment.

The talk provided a comprehensive review of AI techniques across the network protocol stack, highlighting their potential to enhance performance and security. Prof. Chanak also explored key challenges, methodologies, and future opportunities for AI-driven algorithms, protocols, and system designs in next-generation wireless communication.

Distributed Network Tomography: Exact Recovery with Adversarial, Heterogeneous and Sporadic Data

Prof. Gugan Thoppe presented a novel approach to network tomography, focusing on estimating link-level delays from end-to-end path measurements. He examined existing estimation techniques based on data encoding, robust aggregation, and homogenization, highlighting their limitations in practical settings.

To address these challenges, Prof. Thoppe introduced an l₁-minimization-based method that enables exact recovery of heterogeneous linear systems, even in the presence of adversarial behavior and sporadic data availability. The talk also covered convergence guarantees and empirical results, demonstrating the superiority of this approach over existing methods. Extensions to tracking and general optimization were briefly discussed.