Finding Node-disjoint Paths Resilient to Channel Failures in Multi-channel Wireless Networks

This paper focuses on the fundamental problem of finding disjoint paths that are robust and resilient against channel instability while achieving optimal or near-optimal performance. We formulate and analyze the problem of finding k node-disjoint paths resilient to c channel accidents in the sense that if any subset of up to c channels are inoperational, at least one path should still be able to deliver packets. We formulate two versions of the resilient multi-path optimization problem using two performance metrics, the sum of cost and the maximal cost of the k paths. Focusing on directed acyclic graphs, we develop approximation algorithms for both problems. We also develop adapted randomized rounding scheme to further trade off the time and space complexities of our approximation algorithm against the optimality guarantee.