General Convex Agreement with Near-Optimal Communication

A team of researchers including Marc Dufay, Diana Ghinea, and Anton Paramonov has made a significant breakthrough in distributed computing with their paper 'General Convex Agreement with Near-Optimal Communication.' They've developed deterministic synchronous protocols that dramatically improve communication efficiency for Convex Agreement (CA), a Byzantine fault-tolerant consensus problem where outputs must lie within the convex hull of honest parties' inputs. This problem is fundamental to practical applications like robust federated learning and sensor fusion, where honest inputs may differ but should still constrain the final decision. While Byzantine Agreement (BA) has well-established lower bounds, CA protocols have traditionally suffered from higher communication overhead of Θ(Ln²) for L-bit inputs, creating a practical bottleneck for real-world deployment.

The researchers' key innovation lies in using extractor graphs to create deterministic committee assignments resilient against adaptive adversaries, achieving near-optimal communication complexity of O(L·n log n) for finite convexity spaces and O(L·n¹⁺ᵒ⁽¹⁾) for Euclidean spaces ℝᵈ when L = Ω(n·κ), where κ is a security parameter. Their protocols maintain optimal O(n) round complexity and achieve near-optimal resilience t < n/(ω+ε) for constant ε>0, where ω is the Helly number of the convexity space. This represents a substantial improvement over previous approaches and can be leveraged to efficiently solve parallel Byzantine Agreement problems. The work bridges the gap between theoretical lower bounds and practical implementations, making Byzantine-resistant distributed systems more feasible for large-scale applications requiring robust consensus.