HAC-SAGIN: High-altitude computing enabled space–air–ground integrated networks for 6G

The space–air–ground integrated networks (SAGINs) provide a new paradigm for the development of the Internet of Things (IoT) networks by enhancing coverage and deploying computing resources near IoT devices, especially in emergency situations and disaster-hit regions. In the context of the IoT networks, aerial platforms such as unmanned aerial vehicles (UAVs) and high-altitude platforms (HAPs) present in the air layer of SAGINs with access and aerial computing (AC) capabilities have the potential to significantly expand coverage, enhance performance, reduce delay and handle complex computation tasks for IoT devices. Seeking the stated prospect, we propose a high-altitude computing (HAC)-enabled SAGIN leveraging millimeter waves (mmWave) frequency range in which the IoT devices are provided access services by low-earth orbit satellites (LEO-SATs) and HAPs while the HAPs offer AC facility as well. Non-orthogonal multiple access (NOMA) is used as a multiple-access technique with different clustering mechanisms in uplink (UL) and downlink (DL) communication. We aim to establish high-rate data transmission in DL along with minimizing the execution time of IoT devices offloading their data to the HAPs in UL communication. The mmWaves range is targeted to have high-rate data transmissions and NOMA implementation further enhances the bandwidth available for an individual IoT device. For efficient offloading in UL communication, we formulate an optimization problem aiming to minimize the execution time by using the Lagrangian function-based approach. Execution time is minimized by reducing the transmission and computation time, which is attained by the optimization of allocated power and computation resources. Simulation results demonstrate that the proposed HAC-SAGIN is able to establish high-rate transmissions in DL and exhibits a significant decrease in execution time in UL in contrast to the no optimization case. Optimum power assignment improves the achievable rate, leading to reduced transmission times, while optimum core assignment efficiently reduces the computation time. In addition, the offloaded data size-driven NOMA implementation in UL prominently improves the system effective throughput.