Remote and resource-constrained Internet-of-Things (IoT) deployments often lack terrestrial connectivity for task offloading, motivating non-terrestrial networks (NTNs) with onboard multiaccess edge computing (MEC) capabilities. Nevertheless, in the presence of malicious actors, authentication needs to be performed to avoid non-authorized nodes from draining the computing resources of the NTN nodes. As a solution, we propose a four-layer MEC-enabled NTN with unmanned aerial vehicles (UAVs) acting as access nodes, a high altitude platform station (HAPS) acting as coordinator and authenticator, and a constellation of low-Earth orbit satellites (LEOSats) acting as remote MEC servers. We consider a tag-based physical-layer authentication (PLA) scheme to authenticate legitimate users, and formulate a joint task offloading decision and resource allocation for the admitted tasks, which is solved via block coordinate descent. Numerical results show that the PLA scheme is efficient and performs better than the benchmark schemes. We also demonstrate that the proposed scheme is robust against malicious attacks even under relaxed false-alarm constraints.
Internet-of-Things (IoT) devices enable continuous data collection and processing across diverse domains, e.g., Industry 4.0 and smart agriculture. Nonetheless, these devices are typically constrained by limited computing capabilities, limiting their ability to process data locally in a timely manner. To address this limitation, task offloading to remote servers (using multiaccess edge computing (MEC) platforms) may be an alternative to achieve low-latency and secure processing. However, even though there are several benefits in employing MEC platforms, the current coverage provided by terrestrial networks (TNs) is limited to approximately 15% of the Earth's surface. This leaves arge areas, such as remote regions, disaster zones, or conflict areas, without reliable connectivity [1].
To overcome this limitation, non-terrestrial networks (NTNs) have emerged as a promising solution for achieving global coverage by utilizing aerial nodes (e.g., unmanned aerial vehicles (UAVs), high altitude platform station (HAPSs)) and spaceborne assets (e.g., low-Earth orbit satellites (LEOSats)). The dynamic deployment and flexible positioning of these NTN nodes enable seamless service provisioning to IoT devices operating beyond the reach of conventional TN infrastructure [2]. For instance, in [3], the authors explored partial task offloading strategies involving local devices and remote ground servers interconnected via ultra-dense LEOSats backhauls. In the recent work of [4], the authors considered a HAPS as a central entity, and multiple UAVs for coordination between the IoT devices and LEOSats in a multi-layer MEC-enabled NTN, for efficient service provisioning to IoT deployments over large areas. It was shown that the use of multiple layers can underpin a simplified coordination across wide-spanning IoT deployments. Also, the use of intermediate nodes, such as UAVs and HAPS, for task offloading to MEC-enabled LEOSats can provide higher communication rates to the LEOSats than the direct access from the IoT nodes, resulting in reduced transmission delays.
Nonetheless, the same openness and wide-area accessibility that make MEC-enabled NTNs attractive for remote IoT applications also expose the offloading interface to malicious nodes, which could attempt to inject task requests and drain the limited onboard computing and backhaul resources, effectively creating a denial-of-service (DoS) condition for legitimate devices. Thus, it is of utmost importance to authenticate the devices for a proper use of the resources of the NTN. However, conventional cryptographic schemes can be computationally infeasible for constrained IoT devices [5]. To address this challenge, physical-layer authentication (PLA) offers a lightweight alternative by leveraging features of the received signal for rapid verification, making it well-suited for massive IoT access [6]. In the context of NTNs, PLA has been investigated [7], such as in [8], where the authors exploited the Doppler-shift and the received power as features for a PLA scheme to authenticate the LEOSat constellations. Also, in [9], the optimal detection thresholds to differentiate between legitimate and spoofing satellites were evaluated. To the best of our knowledge, existing NTN-PLA studies do not quantify how authentication decisions impact task admission, offloading decisions, and MEC resource allocation in multilayer NTN computing architectures.
Motivated by the above and acknowledging the benefits of multiple NTN layers for IoT-based task offloading, we study secure task offloading in a four-layer MEC-enabled NTN where ground IoT devices intend to offload computation tasks in the presence of malicious nodes. The NTN comprises multiple UAVs that forward the received signals in an amplifyand-forward (AF) manner to a HAPS, which serves as a central coordinator and authenticator and subsequently offloads admitted tasks to MEC-enabled LEOSats. Our key contributions are threefold: (i) we integrate a tag-based PLA mechanism into the uplink offloading pipeline and derive expressions for the test statistic distribution, the probability of false alarm (PFA), the probability of detection (PD), and the optimized threshold under a fixed false-alarm constraint; (ii) we couple the PLA-driven admission outcome with a joint task offloading and LEO computing-resource allocation formulation that prevents unauthenticated tasks from consuming MEC resources; and (iii) we propose an iterative block-coordinate descent solution and demonstrate via simulations that the proposed PLA-enabled framework improves feasible task admission and robustness against malicious offloading compared with benchmarks schemes. To the best of our knowledge, this is the first work that co-designs PLA-based authentication and MEC offloading/resource allocation in multi-layer NTNs.
Notation: Bold uppercase and lowercase letters denote matrices and vectors, respectively; (•) T and (•) H represent the transpose and Hermitian t
This content is AI-processed based on open access ArXiv data.