Towards Optimized Probe Scheduling for Active Measurement Studies

Internet measurement studies often require prolonged probing of remote targets to collect information, yet almost all such studies of which we are aware were undertaken without considering the polluting effects of their unconstrained probing behavior. To help researchers conduct experiments in a more responsible fashion, we present a framework and technique that enables efficient execution of large-scale periodic probing without exceeding pre-set limits on probing rates. Our technique employs a novel scheduling algorithm and leverages knowledge of diurnal traffic patterns to make data collection more efficient (e.g., by probing servers only during periods in which probe results will be most useful). We evaluate our technique in the context of a real-world study and show that it substantially outperforms naı̈ve probing strategies for accomplishing the same goal, sending more probes during useful periods, fewer probes overall, and probing at more precise intervals as required by our measurement applications. Keywords-probe scheduling; responsible network experiments.