Running Time Analysis of Broadcast Consensus Protocols

TL;DR

This paper analyzes the execution time of broadcast consensus protocols (BCPs), showing they can efficiently compute all population protocol predicates and simulate log-space Turing machines within expected interaction bounds.

Contribution

It establishes asymptotically optimal bounds for BCPs' execution time and characterizes polynomial-time BCPs as computing exactly the number predicates in ZPL.

Findings

Expected interactions for population protocol predicates are O(n log n)

BCPs can simulate log-space randomized Turing machines within O(n log n * T) interactions

Polynomial-time BCPs compute exactly the number predicates in ZPL

Abstract

Broadcast consensus protocols (BCPs) are a model of computation, in which anonymous, identical, finite-state agents compute by sending/receiving global broadcasts. BCPs are known to compute all number predicates in $\mathsf{NL}=\mathsf{NSPACE}(\log n)$ where $n$ is the number of agents. They can be considered an extension of the well-established model of population protocols. This paper investigates execution time characteristics of BCPs. We show that every predicate computable by population protocols is computable by a BCP with expected $\mathcal{O}(n \log n)$ interactions, which is asymptotically optimal. We further show that every log-space, randomized Turing machine can be simulated by a BCP with $\mathcal{O}(n \log n \cdot T)$ interactions in expectation, where $T$ is the expected runtime of the Turing machine. This allows us to characterise polynomial-time BCPs as computing exactly the number predicates in $\mathsf{ZPL}$, i.e. predicates decidable by log-space bounded randomised Turing machine with zero-error in expected polynomial time where the input is encoded as unary.