A new approach for reducing end-to-end latency in energy efficient IP-over-WDM networks

Ultra-Reliable Low-Latency Communications (URLLC) services in IP-over-WDM (IPoWDM) backbone networks require stringent end-to-end latency together with energy-efficient operation. Conventional Routing and Wavelength Assignment (RWA) planning primarily minimizes power and may overlook latency inflation introduced by routing and grooming decisions. Hybrid Bypass (HyB) is proposed as a heuristic that jointly targets latency and power through service differentiation: high-priority (latency-sensitive) demands are routed with Direct Bypass (DiB)-like provisioning, while low-priority (best-effort) demands follow a Multi-hop Bypass (MhB)-like strategy and are groomed onto the residual capacity of the virtual topology established by the high-priority stage. A scalable simulator benchmarks HyB against DiB, MhB, and the latency-aware Hottest-first and Comparison (HotC) heuristic. Results show that HyB reduces mean latency by 12.9% for low-priority traffic relative to MhB (up to 37.4% in large-scale topologies) while providing a marginal latency advantage for high-priority traffic relative to DiB (on average 1.5%); overall mean latency is 4 . 75 ms for time-sensitive and 5 . 15 ms for best-effort traffic. Under realistic constraints (finite fibers and wavelength continuity), HyB improves service availability by reducing blocking from ≈ 66 % to < 15 % with minor latency impact, quantifying the latency–availability trade-off that underpins URLLC-oriented planning under constrained resources. Using a component-based power model (router ports, transponders, EDFAs), HyB reduces power by 36.5% for high-priority demands versus DiB and by 57.8% for low-priority demands versus MhB. By integrating selective grooming and service-differentiated routing, HyB provides an energy- and latency-aware RWA planning framework for sustainable, QoS-oriented backbone design. • Hybrid Bypass enables service-differentiated Routing and Wavelength Assignment balancing latency and energy. • Traffic Grooming Rejection Criterion prevents tail-latency inflation. • Direct and Multi-hop Bypass re-implemented with explicit per-demand latency measurements. • Pareto analysis quantifies energy–latency trade-offs across baselines. • Open-source GitHub simulator 1 1 https://github.com/delistavrouk/SimLight . enables reproducible latency–power–blocking benchmarking.