Hopper: Bounded-Memory Collaborative Debiasing for Byzantine-Tolerant Peer Sampling
Researchers propose Hopper, a bounded-memory debiasing protocol that protects Byzantine-tolerant peer sampling from delayed balanced attacks defeating cumulative counters.
The paper shows that frequency-aware debiasing based on cumulative per-identifier counts fails under a delayed balanced attack, where a long benign prefix masks a subsequent adversarial frequency shift. Hopper introduces a bounded-memory debiasing protocol using BitMatcher as the stream estimator, plus BMDecay, a saturation-triggered decay and reconstruction mechanism keeping frequency signal fresh. It also supports authenticated fingerprint-aware trusted collaboration with role-specific debiasing. Experiments show faster recovery from delayed attacks and debiasing comparable to non-debiasing baselines under a fixed memory budget, with a re-identification trade-off at high trusted-node densities.
- Exact cumulative counters fail under delayed balanced attacks in peer sampling.
- Hopper combines BitMatcher estimator with BMDecay saturation-triggered decay.
- Trusted collaboration reduces post-attack pollution but creates re-identification trade-off.
- Occurrence freshness, not exact counting, is key to practical debiasing.
- Products
- HopperBitMatcher
From the original article
1 min read at arxiv.orgByzantine-tolerant peer sampling relies on continuously refreshed views, yet an adversary can bias the identifier streams used to construct them. Frequency-aware debiasing downweights overrepresented identifiers, but existing designs rely on cumulative per-identifier counts.
By Joachim Bruneau-Queyreix · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.