Answers are grounded in this paper and cite their sources.
The rapid growth of the Industrial Internet of Things (IIoT) in hazardous and topologically constrained environments, such as deep-pit mining, requires low-latency, reliable, and spectrum-aware communication and computing infrastructures. In such environments, terrestrial networks often suffer from severe non-line-of-sight (NLoS) blockage, while spectrum reuse with legacy primary systems imposes cognitive coexistence constraints. This paper proposes a simulation-driven optimization framework that integrates an aerial edge machine (AEM) with a smart reflecting panel (SRP) to support autonomous mining robots. Unlike ideal reflecting-surface models that assume lossless unit-modulus phase shifting, the proposed framework adopts a hardware-constrained SRP model whose phase-dependent amplitude response is fitted to measured varactor-tuned element data, and optimizes the reflection coefficients directly on the resulting non-convex manifold. We formulate a worst-case safety-response-time minimization problem by jointly optimizing quasi-static AEM placement, SRP passive beamforming, uplink transmit-power allocation, AEM computation-frequency allocation, and task-partitioning ratios under statistical interference-temperature-limit (ITL) constraints. The problem is solved by a two-level decomposition in which an outer bisection searches the target latency and an inner block coordinate descent (BCD) loop combines closed-form task and computation updates, primal-dual weighted MMSE power allocation, Riemannian manifold optimization (RMO), and a trust-region successive convex approximation (SCA) for placement. Each surrogate step is accepted only if it does not increase a common violation function, so the monotonicity of the inner loop holds by construction; no global optimality is claimed, and the returned latency is achieved by an explicitly verified feasible configuration that upper-bounds the optimum. At a workload of 3.0 Mbits the framework attains 1.05 s against 6.00 s for local execution, an $$82.5%$$ reduction, while maintaining Jain’s fairness index above 0.9 and a fading-averaged cognitive safety headroom of 2.5–12.5 dB below the ITL threshold. The achieved latency lies within 0.05 s of the ideal-reflection bound; the analysis attributes this narrow gap not to near-optimal beamforming but to the deployment being edge-computation-limited, with uplink transmission accounting for roughly $$5.5%$$ of the offloading time, so that the panel’s practical role is insurance against blockage rather than latency reduction. These results delimit the conditions under which hardware-aware and cognition-aware SRP-assisted aerial edge computing benefits mining digital ecosystems and industrial digital-twin deployments.
No takes yet. Share an insight, caveat, or question.
Darem et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: