Energy depletion remains the principal bottleneck constraining the lifetime of Internet of Things (IoT) sensor networks underpinning smart city infrastructures, and most existing clustering and routing protocols address this challenge only partially: Cluster Head (CH) selection is typically governed by a narrow set of criteria, urban-specific propagation effects such as building occlusion are largely overlooked, and performance is most often validated under a single, fixed network configuration, leaving scalability and adaptability across diverse deployment conditions largely unexamined. To close this gap, this paper proposes a novel clustering and routing protocol that dynamically partitions the sensing field into uniform hexagonal clusters, which reconfigure in real time as network area, node count, and spatial distribution change. CH selection is driven by the coati optimization algorithm through a multi-criteria cluster decision function that jointly evaluates sink distance, the Average Scale of Building Occlusions (ASBO), node centrality, and historical CH role count to ensure balanced energy consumption rather than relying on a single dominant metric. Complementing this, a novel relay evaluation function governs multi-hop data forwarding by selecting relay sensor nodes based on their relayrole history, sink proximity, ASBO value, and distance from neighboring CHs, reducing transmission energy costs beyond what direct-to-sink or single-criterion relay schemes achieve. Together, these mechanisms target three coupled optimization objectives: minimizing per-round energy consumption, maximizing network lifetime and throughput, and minimizing end-to-end latency. Extensive simulations evaluate the proposed protocol against its counterparts across varying node densities, sensing areas, and sink locations, rather than a single fixed setup, and show that the proposed protocol extends network lifetime by 184% on average, improves total remaining energy by 147% on average, increases cumulative throughput by 126% on average, and reduces average end-to-end delay by 52% on average relative to the proposed protocol’s counterparts. Also, the proposed protocol achieves an average degradation ratio of only 2.51% across variations in network area size, sink node location, and node density, demonstrating its strong scalability and robustness under diverse IoT network deployment conditions relevant to smart city applications.
Internet of Things (IoT)-enabled electric vertical takeoff and landing (eVTOL) operations require onboard trajectory planning that preserves procedure intent under delayed or missing traffic updates. This study proposes the Procedure-Aware Trajectory Hierarchy with Residual Policy and Safety Shielding (PATH-RS) framework that connects strategic procedure references to onboard tactical planning. Its four-stage procedure model converts assigned procedures into phase-dependent strategic references. Local hazard-aware guidance generates geometric candidates, while recurrent phase-specific residual policies retain temporal context during learned candidate generation. The procedure-aware temporal safety shield selects the executed action through hierarchical candidate screening and fallback. The dual-rate interface separates cached map and procedure information from traffic updates, and an age of information (AoI)-dependent uncertainty margin adjusts finitehorizon separation screening for stale traffic states. Across six scenario families, PATH-RS achieves 95.53% mission success, a 1.27% collision rate, and 100% procedure compliance. Risk-aware model predictive control records 81.07% success and a 16.07% collision rate. Under the evaluated combined delay-and-loss condition, success remains above 97% with complete procedure compliance. In multi-aircraft crossings, distributed cooperation increases pooled success from 60.42% to 98.92% without observed cooperative collisions.
Asynchronous Distributed Key Generation (ADKG) is a distributed key management technique that breaks the global clock limitation and provides a threshold-based key security management scheme for Decentralized Finance (DeFi) threshold wallets. However, the ADKG scheme still suffers from vulnerabilities related to the illegal resale of key shares, compromising the reliability of the threshold signature mechanism relied upon by DeFi threshold wallets. Existing schemes for tracing corrupted nodes that illegally sell key shares are difficult to resist dynamic adversaries in real-world asynchronous network environments. These shortcomings make DeFi fail to achieve consistency, correctness, and non-repudiation of traceability results under asynchronous computation constraints, and to be completely unprotected against dynamic concurrent threats. To address these challenges, we innovatively propose a Traceable, Adaptive, and Packagable Asynchronous Distributed Key Generation (TAP-ADKG) scheme. Its core innovation lies in designing a traceable asynchronous secret sharing algorithm, optimizing the packing polynomial commitment, and constructing a dual-variable asymmetric function to correlate reconstruction and tracing results. This enables black-box tracking of malicious nodes illegally reselling key shares within asynchronous consensus networks. Simultaneously, we propose an adaptive-secure ADKG scheme that eliminates existing adversary corruption assumptions while enabling real-time detection of dynamic adversary eavesdropping in asynchronous networks. Our scheme achieves traceable secret sharing for δ+1 (where δ is the number of corrupted nodes) with an overall communication overhead of O(n2), while maintaining a total computational cost of O(n3)/O(nlogn) for key generation. Simulation results show that even under adaptive Byzantine faults, asynchronous delays, and packet loss, this approach consistently achieves lower end-to-end latency than Bingo and FB-ADKG; by reducing the critical path overhead of asynchronous sharing through CVABA, it maintains excellent overall efficiency.
The Internet of Vehicles (IoV) requires secure and timely authentication for reliable vehicle-to-infrastructure communication. Recent blockchain-based authentication schemes have improved decentralized identity management in IoV systems. However, many of them mainly use blockchain as a storage or auditing layer, while authentication decisions are still performed off-chain and are not strictly synchronized with confirmed on-chain identity states. This may lead to inconsistent identity views among distributed roadside units in highly dynamic environments. To address this gap, this paper proposes a synchronized blockchain-assisted authentication scheme for IoV systems. By binding each authentication attempt to a confirmed onchain authentication-commitment record, the proposed scheme enables roadside units to verify vehicles based on a globally consistent and tamper-resistant identity view. A priority-based transaction packing strategy is further introduced to reduce the confirmation delay of authentication-related transactions. The protocol integrates elliptic curve cryptography, biometric-based local verification, and Physical Unclonable Function (PUF)-enabled roadside units to support mutual authentication and session key establishment. Formal analysis under the Realor-Random model demonstrates session key security, forward secrecy, and user identity privacy. Experimental results show that the proposed protocol requires 28.3456 ms for one authentication session, achieving a 17.65%–39.14% reduction over four of the five compared schemes. It also incurs a communication overhead of 2560 bits, which is 19.19%–47.71% lower than four of the five evaluated schemes. These results demonstrate that the proposed scheme provides an effective balance between security, synchronization, and efficiency, making it suitable for large-scale and highly dynamic IoV authentication scenarios.
In this paper, a novel surrogate modeling method applied to three-dimensional on-chip microwave passive devices for internet of things (IoT) applications, which is based on barycentric form transfer functions combined with artificial neural networks (ANN), is proposed. Among them, this work presents a transfer function parameter extraction algorithm named set support point adaptive Antoulas Anderson (SSP-AAA) algorithm, which avoids the problem of ill-conditioned matrices by optimizing the sequence of support points using a dual criterion based on error and condition number. At the same time, the dimension of the training data was reduced through the set-value strategy. During the model training process, the EM behavior was initially approximated into a transfer function representation through the SSP-AAA algorithm, followed by employing the ANN to establish the mapping relationship between geometric variations and transfer function coefficients. Compared with surrogate modeling based on vector fitting, this method effectively improves the stability of parameter extraction by imposing constraints on the condition number of the Loewner matrix. The test results demonstrate that the proposed method achieves training errors of 2.86%, 1.49% and 2.98% and test errors of 3.54%, 2.06% and 3.36% on three different microwave components, respectively, while maintaining stable coefficient extraction even under wide geometric sampling ranges.
