Adaptive Orbital Direct-to-Device: Rethinking Satellite IoT Architecture

Abstract

Current satellite Direct-to-Device (D2D) architectures often hard-commit to an orbital regime, forcing suboptimal trade-offs between latency, reliability, and energy. We present, to our knowledge, the first adaptive orbital D2D architecture that dynamically switches among LEO, GEO, and hybrid paths based on real-time Doppler, load, and application QoS. Validated via a LoRa satellite-IoT implementation, our approach improves capacity 3× (936 vs. 312 nodes per gateway at 95% PDR), reduces tail latency by 62% (95th percentile: 342ms vs. 892ms), and lowers energy consumption by 31% via retransmission avoidance. Treating orbital choice as a dynamic optimization addresses core limitations of static designs and generalizes beyond LoRa to cellular and NB-IoT D2D systems, charting a path toward resilient, efficient space-terrestrial IoT.