Universal Reliability Bounds for Sparse Networks
Resumen:
Consider a graph with perfect nodes and edges subject to independent random failures with identical probability.The all-terminal reliability (ATR) is the probability that the resulting subgraph is connected. First, we fully characterize uniformly least reliable graphs (ULRG) whose co-rank is not greater than four. Universal reliability bounds are here introduced for those graphs. It is formally proved that ULRG are invariant under bridge-contractions, and maximize the number of bridges among all connected simple graphs with a prescribed number of nodes and edges. A closed-form for the maximum number of bridges is also given, which has an intrinsic interest from a graphtheoretic point of view. Finally, the cost-reliability trade-off is discussed, comparing the number of edges required to reduce the reliability gaps between the least and most reliable graphs. A remarkable conclusion is that the network design is critical under rare event failures, where the reliability-gap between least and most-reliable networks is monotonically increasing with the number of terminals
2021 | |
Agencia Nacional de Investigación e Innovación | |
All-Terminal Reliability Reliability Bounds Uniformly Most Reliable Graphs Uniformly Least Reliable Graphs Ciencias Naturales y Exactas Matemáticas Matemática Aplicada |
|
Inglés | |
Agencia Nacional de Investigación e Innovación | |
REDI | |
https://hdl.handle.net/20.500.12381/648 | |
Acceso abierto | |
Reconocimiento-NoComercial-SinObraDerivada 4.0 Internacional. (CC BY-NC-ND) |
Sumario: | Consider a graph with perfect nodes and edges subject to independent random failures with identical probability.The all-terminal reliability (ATR) is the probability that the resulting subgraph is connected. First, we fully characterize uniformly least reliable graphs (ULRG) whose co-rank is not greater than four. Universal reliability bounds are here introduced for those graphs. It is formally proved that ULRG are invariant under bridge-contractions, and maximize the number of bridges among all connected simple graphs with a prescribed number of nodes and edges. A closed-form for the maximum number of bridges is also given, which has an intrinsic interest from a graphtheoretic point of view. Finally, the cost-reliability trade-off is discussed, comparing the number of edges required to reduce the reliability gaps between the least and most reliable graphs. A remarkable conclusion is that the network design is critical under rare event failures, where the reliability-gap between least and most-reliable networks is monotonically increasing with the number of terminals |
---|