Congestion-aware probe selection for fault detection in networks

Probing is a commonly used technique to identify faults in computer networks. Contrary to the static probes which are otherwise common, we consider the problem of selecting probes in networks which can adapt to changes in traffic dynamics experienced by nodes. This adapting nature of probes reduce the competition of probe induced traffic with other traffic. We formulate a linear programming optimization problem using the total I/O experienced by the node as a cost parameter and show that this minimizes the overall probe cost. Our preliminary simulation experiments conquer this claim.