TITLE:
Telemetry-Aware Risk-Adaptive Routing for Link-Flooding Resilience in Software-Defined Networks
AUTHORS:
Utham Kumar Anugula Sethupathy, Vijayanand Ananthanarayanan
KEYWORDS:
Software-Defined Networking, Link-Flooding Attack, Telemetry, Resilient Routing, Sensitivity Analysis
JOURNAL NAME:
International Journal of Communications, Network and System Sciences,
Vol.19 No.7,
July
31,
2026
ABSTRACT: Adaptive-routing evaluations can overestimate resilience when controllers are assumed to know attack locations. This study presents telemetry-aware risk-adaptive routing (TARA), in which the controller receives only delayed, noisy utilization and delay measurements processed by an exponentially weighted detector with explicit false positives, false negatives, and calibration error. Twenty-four paired synthetic 36-node backbones were evaluated across three attack loads, three detector-quality settings, three baselines, full TARA, and four structural ablations, producing 1728 trials. Under 100% target-capacity attack load and moderate detection, full TARA delivered 97.25% of demand (SD 1.20) versus 93.18% (SD 2.52) for reactive rerouting. The paired improvement was 4.07 percentage points (95% CI 3.39 to 4.74). Its 95th-percentile path latency was 27.15 ms, a paired difference of −6.47 ms (95% CI −7.94 to −5.00) relative to reactive routing. Delivery changed from 97.89% with strong detection to 96.71% with weak detection. Ablations identify path diversity as the dominant source of benefit; risk, overlap control, and utilization prediction produce comparatively modest, condition-dependent differences. These are controlled simulation results, not measurements from a production network.