Event Archive

The early designers of the Internet fostered tremendous innovation by leaving much of the network’s functionality to the programmable computers at its periphery. Unfortunately, the *inside* of the network has been much harder to change. Yet, changing the network is important to make the Internet more reliable, secure, performant, and cost-effective.  The networking research community has struggled for many years to make networks more programmable. What has worked, and what hasn't, and what lessons have we learned along the way?

The Domain Name System (DNS) is the "phonebook" of the Internet, efficiently connecting users to online services. Yet despite its importance, the DNS is also extraordinarily complex and, as a result, fraught with misconfigurations, software implementation bugs, and attack vectors for malicious actors, all of which can impact millions of users. In the past, issues in the DNS have rendered popular services such as GitHub, Twitter, HBO, LinkedIn, Yelp, and Azure inaccessible for extended periods of time. In this talk I will describe our work towards making the DNS as robust as possible.

Encrypted search is the area that focuses on the design, cryptanalysis, implementation of protocols and systems that allow search operations on encrypted data. One aspect of the area that is still poorly understood is its leakage. This talk will go over some recent cryptanalytic results as well as leakage suppression -- a new research direction that focuses on designing “low-leakage” schemes.

Chip power has been steadily increasing with the end of Dennard scaling, requiring ever larger cooling infrastructures and server footprints.  Motivated by this trend, in project Zissou, we are exploring the use of 2-phase liquid immersion cooling for hyperscale public clouds.  Zissou will provide a multi-fold improvement in our IT cooling capability, and in turn unlock innovation across datacenter, server, and software stacks.

In this talk, I am going to introduce Manta, the privacy preserving blockchain platform for DeFi. Manta solves the pseudo-anonymity problem of permissionless blockchain systems by putting privacy guarantee in the design principles. The core technique that Manta uses is zkSNARK, a zero-knowledge proof system that can verify any NP-statement efficiently with zero-knowledge. Currently, Manta has two parts, a decentralized anonymous payment layer that supports minting public tokens to private tokens, and a decentralized anonymous exchange layer that support exchanging tokens using AMM.

We are entering a post-smartphone era, where zillions of new wireless devices compete for the already limited radio spectrum (RF) to obtain wireless connectivity. Addressing the problem of spectrum crunch demands radical new thinking. In this talk, I will present our journey in exploring light as a new spectrum territory, which provides ten thousand times wider bandwidth than RF for wireless communication. Additionally, the physical properties of light offer unique benefits for object tracking and behavioral sensing.

DUKE-FUQUA DECISION SCIENCES SEMINAR

Agreement protocols for partially synchronous or asynchronous networks tolerate fewer than one-third Byzantine faults. If parties are equipped with trusted hardware that prevents equivocation, then fault tolerance can be improved to fewer than one-half Byzantine faults, but typically at the cost of increased communication complexity. In this work, we present results that use small trusted hardware without worsening communication complexity assuming the adversary controls a fraction of the network that is less than one-half.

Facebook is one of the biggest Internet content providers in the world. In this talk, we will take a sneak peek of how the edge, backbone, and datacenter networks behind the scene that supports Facebook’s massive infrastructure, which in turn supports its more than 2 billion users through Facebook, Instagram, Whatsapp, Messenger, Oculus and many more applications.

Many attacks have shown that deep learning models trained on private data of users can leak sensitive information of the users. Differential Privacy is a provable way to prevent such attacks. However, training deep learning models using DP introduces several new challenges both in terms of privacy vs accuracy tradeoffs and in the resource cost of the process. In this talk, I will highlight some of the problems we encountered, our solutions for resolving them and mention many important open problems.

Duke alum Cade Metz will discuss his new book, Genius Makers: The Mavericks Who Brought AI to Google, Facebook, and the World, a sweeping look at the rise of "AI" (Artificial Intelligence) over the last 10 years -- and the decades of history that made this possible. He will also discuss his time at Duke and his path to becoming a New York Times reporter. He will be interviewed by North Carolina State professor Ross Bassett, before taking any and all questions from students.

The information security community has devoted substantial effort to the design, development, and universal deployment of strong encryption schemes that withstand search and seizure by computationally-powerful nation-state adversaries. In response, governments are increasingly turning to a different tactic: issuing subpoenas that compel people to decrypt devices themselves, under the penalty of contempt of court if they do not comply.

The talk mainly focuses on two topics. The first topic is bandit learning for Bounded Mean Oscillation (BMO) functions, where the goal is to "maximize'' a function that may go to infinity in parts of the space. For an unknown BMO function, I will present algorithms that efficiently finds regions with high function values.

Network attacks are on the rise, and many of them can be traced to a common root cause---the Internet does not have security support in its architecture. Existing proposals either need to make intrusive changes to the Internet, or resort to bolt-on protection for each discovered attack. In the Poise (Programmable In-network Security) project, we are rethinking how to develop a secure foundation for the next-generation Internet. Poise leverages technological advances in emerging programmable networking hardware, and it takes a three-step approach.

Recently, Merkle trees have been proposed as a way to scale block validation in cryptocurrencies such as Ethereum (or Bitcoin). The key insight is that block validators can verify Merkle proofs of account balances (or of unspent coins) against a Merkle root of the cryptocurrency’s state (i.e., the database of every user’s balance). This so-called stateless validation approach eliminates the need for block validators (e.g., miners, P2P nodes) to store large amounts of data and access it from disk during validation, which can be slow.

How can we efficiently represent and enumerate the results of theta-join queries? Factorization techniques have found notable success as a compact representation scheme that allows for enumeration algorithms where the results are produced incrementally after a preprocessing phase. However, these factorized representations have mainly been limited so far to equi-joins. This talk will present factorization techniques for general theta-join queries where the join conditions between relations go beyond equality.