A Long-Short-Term Stress-Tolerant Routing Algorithm for LEO Satellite Network

Abstract

As multidomain networks integrate, low Earth orbit (LEO) satellite network embraces more devices with direct-to-device communication. However, the surge in devices brings heavy traffic stress to LEO satellite network, which has posed a great challenge to routing path planning and on-board resource allocation. This article proposes a long-short-term (LST) stress-tolerant routing algorithm for the two-time-scale cooperation of path planning and power allocation. To balance the global stress distribution with low signaling overhead, its long-term (long update period) algorithm uses a delay-aware artificial potential field to reconstruct the topological costs, allowing an enhanced A-star to efficiently plan more flexible paths. To regulate the local stress trend, the short-term (short update period) algorithm optimizes the transmit power and reduces the queueing delay based on channel and queue states. It uses the best square approximation to make a lower burden computation. In addition, the long- and short-term algorithms establish bidirectional feedback with the nodes' communication willingness. In numeral results, LST raises the critical tolerance stress and ensures stable data switch under high stress. Moreover, it guarantees dependability in cases of bursty traffic and unstable intersatellite links.