Conference Brief — 2026-10-20

In brief

Privacy and Metadata Protection: New work introduces Onion Franking for abuse reporting in mix-based messaging systems, and Optimix offers a solution for latency optimization in modern mixnets.
Microarchitectural Mitigations: The Oreo system is proposed to protect ASLR against bypass attacks via side channels.
Scalable Consensus: Pallas and Aegis provide rollback resilience in TEE-aided blockchain consensus, while Pando introduces a highly scalable BFT protocol based on committee sampling.

Introduces Onion Franking, a mechanism for abuse reporting in metadata-hiding messaging systems that rely on onion encryption or mixnets.
Demonstrates that breaking with abstractions used in prior work (such as compactly committing authenticated encryption) and relying directly on underlying cryptographic tools enables savings in communication costs.
Achieves over an order of magnitude reduction in computation overhead compared to the most efficient prior work applicable to onion encryption-based systems.
Matthew Gregoire, Margaret Pierce, Saba Eskandarian. “Onion Franking: Abuse Reports for Mix-Based Private Messaging.” NDSS — https://ndss-symposium.org/ndss-paper/onion-franking-abuse-reports-for-mix-based-private-messaging

Presents OptiMix, an approach for latency optimization in modern mixnets that generalizes across cascade and stratified topologies.
Evaluates routing strategies through simulations, finding that applying cover routing generation methods results in higher entropy and fewer fully compromised paths.
Analyzes mixnode adversaries and measures latency and geographical data using datasets like Nym and RIPE.
Mahdi Rahimi. “OptiMix: Scalable and Distributed Approaches for Latency Optimization in Modern Mixnets.” NDSS — https://ndss-symposium.org/ndss-paper/optimix-scalable-and-distributed-approaches-for-latency-optimization-in-modern-mixnets

Proposes Oreo, a defense against microarchitectural side channel attacks that bypass Address Space Layout Randomization (ASLR).
Implements microarchitecture changes and evaluates them across various benchmarks to measure latency overheads and execution times.
Observes latency overhead variations in LEBench ranging from -6.39% to +6.16% compared to insecure baselines.
Shixin Song, Joseph Zhang, Mengjia Yan. “Oreo: Protecting ASLR Against Microarchitectural Attacks.” NDSS — https://ndss-symposium.org/ndss-paper/oreo-protecting-aslr-against-microarchitectural-attacks

Develops OSAVRoute, a non-cooperative measurement system to accurately detect the deployment of Outbound Source Address Validation (OSAV).
Demonstrates that OSAVRoute achieves an accuracy of 99.4% and assisted a nationwide ISP in correcting their OSAV misconfigurations.
Measures 754 addresses across 510 /24 prefixes within the ISP network, finding that 10.3% of tested addresses lack OSAV deployment.
Shuai Wang, Ruifeng Li, Li Chen, Dan Li, Lancheng Qin, Qian Cao. “OSAVRoute: Advancing Outbound Source Address Validation Deployment Detection with Non-Cooperative Measurement.” NDSS — https://ndss-symposium.org/ndss-paper/osavroute-advancing-outbound-source-address-validation-deployment-detection-with-non-cooperative-measurement

Addresses the problem of privacy-preserving attribute-driven community search over attributed graphs for data-driven applications.
Proposes methods to find high-quality subgraphs meeting specific attribute requirements without compromising data privacy.
Evaluates the approach by selecting attributes for communities and ensuring attribute coverage across community vertices.
Fangyuan Sun, Yaxi Yang, Jia Yu, Jianying Zhou. “PACS: Privacy-Preserving Attribute-Driven Community Search over Attributed Graphs.” NDSS — https://ndss-symposium.org/ndss-paper/pacs-privacy-preserving-attribute-driven-community-search-over-attributed-graphs

Introduces Paladin, a defense against LLM-generated phishing emails that are free from spelling mistakes and use topic-specific content.
Employs a trigger-tag paradigm, demonstrating over 85% phishing detection accuracy across three insertion strategies and four LoRA rank settings.
Achieves a 5% improvement in detection accuracy and time over baseline methods in evaluated scenarios.
Yan Pang, Wenlong Meng, Xiaojing Liao. “Paladin: Defending LLM-enabled Phishing Emails with a New Trigger-Tag Paradigm.” NDSS — https://ndss-symposium.org/ndss-paper/paladin-defending-llm-enabled-phishing-emails-with-a-new-trigger-tag-paradigm

Presents Pallas and Aegis to provide rollback resilience in Trusted Execution Environment (TEE)-aided blockchain consensus protocols.
Demonstrates through experimentation on real-world infrastructure that Pallas delivers high performance with negligible overhead in stable conditions.
Outperforms state-of-the-art BFT consensus systems by up to 41% in throughput and 29% in latency overhead.
Jérémie Decouchant, David Kozhaya. “Pallas and Aegis: Rollback Resilience in TEE-Aided Blockchain Consensus.” NDSS — https://ndss-symposium.org/ndss-paper/pallas-and-aegis-rollback-resilience-in-tee-aided-blockchain-consensus

Develops Pando, a highly scalable Byzantine Fault-Tolerant (BFT) protocol based on committee sampling to handle large numbers of replicas.
Shows that for the same desirable probability of safety and liveness violation, Pando uses a committee size an order of magnitude lower than existing works.
Assesses CPU and bandwidth usage, illustrating why Pando outperforms existing protocols even with large committee sizes.
Xin Wang, Haochen Wang, Haibin Zhang, Sisi Duan. “Pando: Extremely Scalable BFT Based on Committee Sampling.” NDSS — https://ndss-symposium.org/ndss-paper/pando-extremely-scalable-bft-based-on-committee-sampling