internet of things

To address environmental monitoring requirements in greenhouse and red-house scenarios, this paper designs and implements a master–slave IoT system based on the STM32F103C8T6 microcontroller (ARM Cortex-M3). The system integrates DHT11 (temperature/humidity), YL-69 (soil moisture), FC-37 (raindrop detection), HC-SR04 (obstacle distance), and SSD1306 OLED for real-time multimodal sensing. Inter-node synchronization is achieved via HC-05 Bluetooth with sub-200 ms latency. A lightweight embedded coordination mechanism enables millisecond-scale sampling and threshold-triggered alerts with <300 ms response time. The system achieves ultra-low standby current of 15 μA and sustains stable operation for over 72 hours. This work presents the first STM32-based dual-node Bluetooth-synchronized architecture for environmental monitoring, empirically demonstrating the Cortex-M3 core’s superior energy efficiency and peripheral integration capability compared to AVR microcontrollers.

This study systematically evaluates whether Rust can compete with C in performance and resource efficiency for microcontroller firmware development and assesses its industrial viability. Two teams independently implemented identical industrial IoT firmware—one in Rust and the other in C—and key metrics including development effort, memory footprint, and execution speed were compared on real hardware. This work presents the first systematic comparison of the two languages in a genuine industrial context and introduces Ariel OS, a lightweight Rust-based runtime. Empirical results demonstrate that Rust matches or exceeds C in both resource utilization and execution performance, while Ariel OS exhibits a smaller binary footprint, collectively establishing Rust as a reliable and competitive choice for microcontroller firmware development.

To address the challenges of bandwidth constraints, high energy consumption, and inability to support full-package transmission in firmware over-the-air (OTA) updates for low-power wide-area networks (LPWANs) such as LoRaWAN, this paper proposes a lightweight, incremental OTA update mechanism tailored for resource-constrained IoT end-devices. The method introduces an embedded-friendly binary differencing algorithm that efficiently generates highly compressed delta patches on-device, coupled with LoRaWAN-aware adaptive fragmentation and robust integrity verification to ensure end-to-end reliable firmware reconstruction. Experimental results demonstrate that, compared to full-image updates, the proposed approach reduces transmission volume by over 70%, shortens update latency and energy consumption by more than 60%, and significantly extends the maintenance-free operational lifetime of battery-powered nodes. This work provides a practical and sustainable solution for firmware evolution in LPWAN environments.

This work addresses the challenge of end-to-end machine learning inference on microcontroller-class edge devices under stringent constraints on memory, energy consumption, and latency. To bridge the gap between conventional machine learning pipelines and embedded deployment realities, the authors propose a robust design framework tailored for resource-constrained environments, encompassing data acquisition, preprocessing, model compression, and streaming deployment. The framework integrates sampling buffers, feature dimensionality reduction techniques (e.g., RMS, spectral features, MFCCs), validation strategies for class imbalance, and co-optimization of models with runtime systems to form a complete embedded ML pipeline. Experimental evaluations on two representative tasks—inertial human activity recognition and keyword spotting—demonstrate that the proposed approach enables efficient, practical, and robust on-device inference, significantly narrowing the divide between general-purpose machine learning methodologies and embedded implementation requirements.

Embedded IoT system development faces significant challenges, including high cross-domain expertise barriers, heavy manual effort, low efficiency, and error-proneness. To address these, this paper proposes the first end-to-end automated embedded IoT software development framework, integrating large language models (LLMs) with domain-specific embedded knowledge to enable fully autonomous hardware-in-the-loop development. Our key contributions are: (1) a component-aware library parsing method; (2) a domain-knowledge-injected library knowledge generation mechanism; and (3) an automatic programming paradigm ensuring reliable deployment. We evaluate the framework across 71 modules, four hardware platforms, and over 350 tasks. Results show a code accuracy of 95.7% and an end-to-end task success rate of 86.5%, outperforming human experts by up to 53.4% in task completion.

This work addresses the challenge that existing AI methods struggle to jointly model the tight coupling between software logic and physical hardware behavior in hardware-in-the-loop (HIL) development of embedded and IoT systems, often leading to deployment failures. To tackle this, the authors propose a skill-oriented agent architecture tailored for HIL scenarios and introduce IoT-SkillsBench, a novel real-hardware evaluation benchmark. The framework systematically assesses AI agents across multiple platforms, peripherals, and task complexities through three agent configurations enhanced by skill augmentation, structured expert knowledge injection, and real-hardware validation. Experimental results demonstrate that, over 378 real-world deployments, agents equipped with human-expert-derived skills achieve near-perfect cross-platform task success rates, substantially outperforming baseline approaches.

This study addresses the challenges in testing Internet of Things (IoT) software, particularly the complexity of external dependencies and insufficient test case effectiveness, which have lacked systematic empirical investigation. It presents the first large-scale analysis of testing practices in open-source IoT projects, integrating assessments of test effectiveness, categorization of testing challenges, and mining of mock usage patterns. The findings reveal that despite the substantial volume of tests, their effectiveness is generally limited, with managing external dependencies emerging as a central difficulty. Moreover, the judicious application of mock objects significantly enhances test coverage and quality. This work establishes the first empirical benchmark for IoT software testing and offers concrete directions for improving testing practices in this domain.

This study addresses the insufficient cybersecurity robustness of LoRaWAN-based smart lighting systems in real-world deployments. We propose a practical attack-testing framework integrating controlled lab experiments with iterative field trials to systematically evaluate security vulnerabilities in commercial off-the-shelf devices. Our key contributions include: (i) identifying high-power, short-range physical-layer interference as the predominant threat to end-devices and gateways—previously underexplored; and (ii) the first empirical validation that gateway redundancy significantly enhances system resilience against such interference. Experimental results demonstrate that most protocol-layer attacks fail in operational environments, whereas physical-layer jamming remains a tangible threat; our redundancy mechanism improves communication availability by over 70% under strong interference. The work establishes a reusable, empirically grounded evaluation methodology and deployable mitigation strategies for low-power wide-area IoT security design.

This study addresses the lack of systematic empirical comparisons between Zigbee and Matter over Thread under real-world network conditions. Leveraging commercial off-the-shelf hardware, we construct a multi-hop testbed and conduct experiments involving load stress, fault injection, and scalability to quantitatively evaluate the trade-offs between agility, efficiency, and scalability in realistic multi-hop scenarios. Our results reveal that Zigbee achieves lower latency and reduced overhead in small-scale static networks, whereas Matter over Thread demonstrates superior scalability and more stable throughput in dynamic, multi-hop environments. This work provides critical empirical evidence to inform protocol selection for Internet of Things deployments.

This work addresses the challenge of enabling efficient autonomous perception on edge Internet-of-Things (IoT) devices under stringent energy, memory, and latency constraints by proposing EdgeSpike, a novel framework that integrates hybrid surrogate gradient and direct encoding training, hardware-aware neural architecture search, event-driven runtime execution, a custom sparse SIMD core, and lightweight local plasticity rules. For the first time in real-world IoT scenarios, EdgeSpike achieves continuous online learning without backpropagation. Evaluated across five sensing tasks, it attains an average accuracy of 91.4%—comparable to INT8 CNNs—while reducing inference energy consumption by 4.6× to 47× and maintaining end-to-end latency at or below 9.4 ms. A 64-node field deployment demonstrates a 6.3× extension in battery life, with only a 0.7 percentage point accuracy drop under seasonal environmental shifts.