Dr. Stefanie Roos is an assistant professor for distributed systems at TU Delft and the Delft Blockchain Lab. Her work deals with trade-offs between privacy, security, and performance in decentralized systems. She contributed to the censorship-resistant P2P network Freenet and designed SpeedyMurmurs, a routing algorithm for payment channel networks like Lightning. Her current research is focused on improving layer2 protocols for blockchains as well as designing more efficient anonymity systems.
Before joining TU Delft, she was a post-doctoral researcher at University of Waterloo, working with Prof. Ian Goldberg, and a PhD scholar at TU Dresden and TU Darmstadt. Her PhD thesis, supervised by Prof. Thorsten Strufe, won the KuVS award for the best PhD thesis in the area of networks and distributed systems in Germany.
Some examples of my work:
- Recent survey on layer2/off-chain transactions
- SpeedyMurmurs: Settling Payments Fast and Private
- Talk on Routing Off-Chain Transactions at the MIT Bitcoin Expo
- Anonymous addresses for efficient and resilient routing in F2F overlays
- Measuring freenet in the wild: Censorship-resilience under observation
- PhD thesis: Analyzing and Enhancing Routing Protocols for Friend-to-Friend Overlays
See my Google scholar page for a full list of publictions.
Courses I am/was involved in
CSE1405 Computer Networks
CS4160 Blockchain Engineering
CS4215 Quantitative Performance Evaluation for Computing Systems
At the moment, I offer topics from three areas of research:
- Fundamental research on scalability of blockchains, in particular payment channel networks,
- Improving anonymity networks, in particular Tor, and
- Application of decentralized infrastructures such as blockchain to real-world problems (in collaboration with TNO and Syscoin).
The details of the topic are to be worked out by the student, so please see the research questions below as initial examples.
Fundamental Research on Scalability of Blockchains
Payment channel networks such as Lightning enable fast, low-cost, and energy-efficient transactions and can drastically increase the throughput of blockchains. Parties utilize the blockchain to lock collateral and then can use this collateral to conduct transactions without initiating the global consensus of the blockchain for every transactions (See here for a survey of the topic). Some open research questions include:
- How can we incentivize locking collaterals in payment channel networks to increase their liquidity?
- How do failures and denial-of-services attacks, which reduce the probability of successfully conducting a payment, affect the payment success and how can the impact of such failures/attacks be mitigated?
- How does routing a payment via multiple intermediary nodes in a payment channel network affect the privacy of the involved parties? How can we maximize the privacy without sacrificing efficiency?
Improving Anonymity Networks
Given large-scale surveillance and the persecution of activists, anonymity network play an important part in protecting freedom of speech and human rights. However, the most famous anonymity network Tor suffers from low performance and network layer anonymity in recent applications such as blockchain is essentially non-existent. (See here for some background on Tor, here for anonymity of Bitcoin, here for an overview of anonymization techniques)
- How can we best integrate congestion control into Tor to improve speed without reducing privacy and security?
- Is QUIC a more suitable transport layer protocol for Tor?
- Mix networks provide performance-anonymity trade-offs that are suitable for blockchains. How can we integrate them into existing protocols?
- If an application offers anonymity, should anonymity be compulsory for all users, opt-in, or opt-out? What are the impact of the different strategies on privacy and performance?
Application of Decentralized Architectures in Real-world Systems
- (With TNO) Can we extend approaches for decentralized storage such that multi-facetted storages requirements (such as storing content only on participants from the EU) are accounted for?
- (With Syscoin) How can payment channel networks increase the efficiency of trading digital assets in Syscoin?
- (With Syscoin) How can zero-knowledge proofs be used to realize two-way sidechains, i.e., blockchains that can transfer between multiple currencies?