The simultaneous failure problem motivates us to propose the RBPS approach to select a proper one-hop backup path for any node pair.
The RBPS approach is built upon two widely used assumptions: 1) the delay of [Path.
Based on the grid quorum system and traceroute probing, we propose a 3-phases RBPS approach, including link-state probing, link-state table distribution and recommendation of the backup path.
The communication overhead of the RBPS approach is bound by 0([n.
Identifying the best one-hop backup path in RBPS makes every node incur N + 4 [square root of N-3] messages sizing of m/4 (N - 1)([square root of N - 1]) + 8(N - 1) + 16([square root of N - 1]) bytes in total, where m denotes the bit length of the Bloom filter and N denotes the number of nodes in the overlay network.
Besides, the effectiveness of our RBPS approach relies on the design of the Bloom filter that encodes the IP address sequence on the default path.
This paper proposes the RBPS approach to improve the robustness of the selected one-hop path, and further employs the Bloom filter to considerably reduce the communication overhead.
Optimized parameters of the Bloom filter can improve the performance of our RBPS approach.
Extensive simulations are conducted to evaluate the simultaneous failure probability in the RBPS approach.
When such a number is beyond 40, the simultaneous failure probability in our RBPS approach is much smaller than that in the SAP approach.
The backup paths selected by the RBPS approach exhibit lower simultaneous failure probability.
9(c), our RBPS approach identifies a more robust backup path that can detour around the failed links on the default path.