FROOT: Future-Proof, Trustworthy Telemetry

With the growth of the Internet and its importance in supporting the US economy, business, health, education and other services, it is critical to ensure both high performance and high availability of the networks underlying it. Increasingly, such networks include a heterogeneous set of network switches and other devices which must be monitored and controlled in a coordinated manner. Emerging networked applications, such as cloud gaming or cloud streamed augmented reality, are expected to further stress both control systems and network monitoring by requiring real-time response to rapid changes in traffic workloads. This project aims to address the needs of future network control by enabling a network telemetry infrastructure that can provide timely, accurate, and trusted information about ongoing activities in the network.

ONSET: Optics-enabled Network Defenses for Extreme Terabit DDoS Attacks

Distributed Denial of Service (DDoS) attacks continue to present a clear and imminent danger to critical network infrastructures. DDoS attacks have increased in sophistication with advanced strategies to continuously adapt (e.g., changing threat postures dynamically) and induce collateral damage (i.e., higher latency and loss for legitimate traffic). Furthermore, advanced attacks may also employ reconnaissance (e.g., mapping the network to find bottleneck links) to target the network infrastructure itself. In light of these trends, state-of-art defenses (e.g., advanced scrubbing, emerging software-defined defenses, and programmable switching hardware) have fundamental shortcomings. This project will develop a new framework, referred to as “Optics-enabled In-Network defenSe for Extreme Terabit DDoS attacks” (ONSET). The framework makes a case for new dimensions of defense agility that can programmatically control the topology of the network (in addition to the processing behavior) to tackle advanced and future attacks. The project will facilitate the use of optical technologies as an exciting visual medium for engaging K-12 students via suitable channels for dissemination. The project will also result in new course materials at the intersection of optical networking, software-defined networking, and network security to enable students to become domain experts in this emerging problem space.

AI for Cloud Ops

Today’s Continuous Integration/Continuous Development (CI/CD) trends encourage rapid design of software using a wide range of customized, off-the-shelf, and legacy software components, followed by frequent updates that are immediately deployed on the cloud. Altogether, this component diversity and breakneck pace of development amplify the difficulty in identifying, localizing, or fixing problems related to performance, resilience, and security. Existing approaches that rely on human experts have limited applicability to modern CI/CD processes, as they are fragile, costly, and often not scalable.

Enabling Intelligent In-Network Computing for Cloud Systems

With the network infrastructure becoming highly programmable, it is time to rethink the role of networks in the cloud computing landscape beyond just packet delivery. The network itself emerges as a computing platform with a unique advantage of full network visibility. This project enables advanced approximate telemetry (e.g., sketches) with relevant applications on programmable networks (e.g., programmable switches, SmartNICs, and FPGAs) for cloud system management. Specifically, we will develop a cloud-native, approximate telemetry framework to offer low-overhead, fine-grained, real-time visibility into the underlying network traffic. Moving forward, this line of research with “intelligent” networking aims to provide insights and system abstractions to improve the performance, reliability, and security of cloud systems.

Privacy-Preserving, Automated Data Sharing Framework

Our conversations with leading cloud and AI vendors across market verticals (e.g., security, telemetry, finance) tell us that at every step along the way, lack of access to realistic and diverse data from multiple deployments hampers innovation. For instance, data-driven products trained on data not representative of real customer environment, there is no way to quantitatively assess products; machine learning workflows experiences data drift, and product feedback is not quantitative. The result today is poor products, lack of transparency, lots of effort in debugging/reproduction/resolution, and impossibility to share insights across diverse customers.

Scalable Learning from Distributed Data for Wireless Network Management

The transition to 5G is expected to witness not only an emergence of new applications such as mobile augmented and virtual reality, but also opens up the attack surface to both known, and previously unknown threats. Thus, wireless networks of the future will need better control and management at different temporal and traffic aggregation granularities (e.g., how to allocate spectrum, how to quarantine distributed attacks etc.). This project aims to develop scalable, machine learning based analytics on the data from a large set of geographically distributed wireless core network entities such as base stations. The research will enable new approaches for: (a) compressing the raw data via novel summaries and sketches, that reduce overhead while simultaneously enabling highly accurate scalable analytics (b) scalable yet highly flexible distributed learning approaches that are built upon the emerging federated learning paradigm and (c) flexible allocation of bandwidth to support the control plane analytics that minimizes the impact on the data plane.