Core Concepts
Three scattered agents can localize a black hole in a dynamic 1-interval connected ring in Θ(n^2) moves and rounds, which is optimal.
Abstract
The paper investigates the problem of searching for a black hole (a node that silently destroys any visiting agent) in a dynamic 1-interval connected ring by a set of scattered agents (i.e., agents starting from arbitrary locations).
The key insights are:
In the endogenous communication models (where agents can only communicate when co-located), the black hole search problem is unsolvable using three scattered agents, in contrast to the colocated case where it is solvable.
To circumvent this impossibility, the authors consider exogenous communication models where agents can use pebbles or whiteboards.
They show that any optimal size algorithm (using three agents) for black hole search in dynamic rings requires Ω(n^2) moves and rounds in the whiteboard model. This highlights the significant complexity increase compared to the static case, where two agents can solve the problem in O(n) moves and rounds.
The authors provide a tight algorithm that solves the problem in O(n^2) moves and rounds using three agents in the pebble model.
The paper provides a comprehensive analysis of the computational complexity of black hole search by scattered agents in dynamic rings, identifying both impossibility results and optimal algorithms.
Stats
Any algorithm solving the black hole search problem with three agents requires Ω(n^2) moves and Ω(n^2) rounds.
Quotes
"To the best of our knowledge this is the first paper examining the problem of searching a black hole in a dynamic environment with scattered agents."
"We show that any optimal size algorithm solving Bhs requires Ω(n^2) moves and Ω(n^2) rounds in the whiteboard model."
"Finally, our lower bound is tight: we provide an algorithm that solves Bhs in the pebble model in O(n^2) rounds and moves using three agents."