TRACK: Traffic Routing and Allocation with Capacity-aware KL-mapping in LEO Satellite Networks

초록

This work studies joint routing and virtual network function (VNF) maaping for service function chaining in software defined networks (SDN)-enabled low earth orbit (LEO) satellite networks. Each satellite hosts pre-loaded VNFs but has limited CPU, memory, and link capacity, making routing and mapping tightly coupled. We propose traffic routing and allocation with capacity-aware KL-mapping (TRACK), a two-stage method that selects congestion-aware feasible routes and maps VNFs by matching each satellite's residual resources to VNF demand via Kullback-Leibler (KL) divergence. On a 66-satellite Iridium topology, TRACK improves the service success rate by up to 18.42% over random mapping and 10.28% over greedy mapping by alleviating resource imbalance and preventing saturation. © 2026 IEEE.

키워드