Wireless IoT Research Articles

Impact of Embedded Deep Learning Optimizations for Inference in Wireless IoT Use Cases

Wireless internet-of-things devices typically transmit data to high-end computing platforms such as edge devices or cloud devices for further processing. However, processing the data at the edge or even on the constrained embedded devices using neural network can eliminate the need for high-throughput links and provides several benefits in terms of latency, reliability, privacy and energy consumption. In this article, we quantify how efficient neural networks can run on embedded devices for three typical wireless use cases. To this end, we give an overview of different optimizations and strategies for embedded deep learning inference on constrained devices. Next, we quantify the performance impact of optimized neural networks for edge and embedded inference, which perform up to 2.5x and 20x faster and consume 20x less energy at the cost of less than 2 percent accuracy difference for classification models. Although most published oversized models cannot run on typical embedded devices, with optimizations, we achieve efficient embedded inference and mitigate the need for raw data transmissions and thus preserve privacy. Finally, we discuss trends found in embedded deep learning use cases and present insights between design- and run-time metrics to predict model memory, storage and energy consumption together with model inference time.

Open Issues for Intelligent Connectivity Wireless Sensor Networks (WSNs) and IoT: State of the Art

Open Issues for Intelligent Connectivity Wireless Sensor Networks (WSNs) and IoT: State of the Art

Building Artificial-Intelligence Digital Fire (AID-Fire) system: A real-scale demonstration

The identification of building fire evolution in real-time is of great significance for firefighting, evacuation, and rescue. This work proposed a novel framework of Artificial-Intelligence Digital Fire (AID-Fire) that can identify complex building fire information in real-time. The smart system consists of four main parts, Internet of Things sensor network (data collection and transfer), cloud server (data storage and management), AI Engine (data processing), and User Interface (fire information display). A large numerical database, containing 533 fire scenarios with varying fire size, positions, and number of fire sources, is established to train a Convolutional Long-Short Term Memory (Conv-LSTM) neural network. The proposed fire digital twin is demonstrated and validated in a full-scale fire test room (26 m 2 ). Results show that the AI engine successfully identify the fire information by learning the spatial-temporal features of the temperature data with a relative error of less than 15% and a delay time of less than 1 s. Moreover, detailed fire development and spread can be accurately displayed in the digital-twin interface. This proposed AID-Fire system can provide valuable support for smart firefighting practices, thus paving the way for a fire-resilient smart city. • A smart building digital twin was developed to identify fire scene in real-time. • Wireless IoT sensors and deep learning model drive the digital twin of building fire. • AI model identifies fire information in digital twin with an error <15% and time lag <1 s. • The digital fire system demonstrates a prototype of future smart firefighting system.

920MHz Path Loss Prediction Formula Based on FDTD Method for IoT Wireless System close to Ceiling with Concrete Beam

920MHz Path Loss Prediction Formula Based on FDTD Method for IoT Wireless System close to Ceiling with Concrete Beam

Treatment and Disposal of Other Medical Wastes based on Bluetooth and LoRa Wireless IoT

With the maturity of the Internet of Things technology, medical institutions have established a positioning and sensing wireless IoT platform for the entire medical institution by using technologies such as Bluetooth positioning beacons, LoRa communication base stations, positioning and map engines. The intelligent medical waste collection and transfer vehicle integrates the wireless Internet of Things platform in the hospital based on 4G/5G modules, Bluetooth positioning terminals, electronic weighing and other technologies. An information-based, intelligent, and visualized "smart medical waste" management system covering all aspects of medical waste generation, collection, temporary storage, transfer, and disposal has been established in medical institutions. This set of solutions can greatly reduce the number of communication base stations deployed in hospital indoor buildings, the number of weak current construction and access switches in hospitals, save a lot of unnecessary costs for weak current construction and access switches and meet the real-time needs of intelligent medical waste collection and transfer vehicles such as real-time positioning, real-time tracking, video linkage, trajectory playback, and transfer route deviation alarms, which have certain reference significance in the practice of intelligent medical waste construction technology in hospitals based on smart IoT. The article describes one case of implementation of the technology in a university Hospital.

End-to-End Image Classification and Compression With Variational Autoencoders

The past decade has witnessed the rising dominance of deep learning and artificial intelligence in a wide range of applications. In particular, the ocean of wireless smartphones and IoT devices continue to fuel the tremendous growth of edge/cloud-based machine learning (ML) systems, including image/speech recognition and classification. To overcome the infrastructural barrier of limited network bandwidth in cloud ML, existing solutions have mainly relied on traditional compression codecs such as JPEG that were historically engineered for human-end users instead of ML algorithms. Traditional codecs do not necessarily preserve features important to ML algorithms under limited bandwidth, leading to potentially inferior performance. This work investigates application-driven optimization of programmable commercial codec settings for networked learning tasks such as image classification. Based on the foundation of variational autoencoders (VAEs), we develop an end-to-end networked learning framework by jointly optimizing the codec and classifier without reconstructing images for a given data rate (bandwidth). Compared with the standard JPEG codec, the proposed VAE joint compression and classification framework achieves classification accuracy improvement by over 10% and 4%, respectively, for CIFAR-10 and ImageNet-1k data sets at data rate of 0.8 bpp. Our proposed VAE-based models show 65%–99% reductions in encoder size, <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\times 1.5$ </tex-math></inline-formula> – <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\times 13.1$ </tex-math></inline-formula> improvements in inference speed, and 25%–99% savings in power compared to baseline models. We further show that a simple decoder can reconstruct images with sufficient quality without compromising classification accuracy.

A Survey of Data Aggregation Protocols for Energy Conservation in WSN and IoT

Conservation of energy has been a major concern for Wireless Sensor Networks (WSNs) and IoT applications. Several strategies were devised, aimed at optimizing energy consumption in these applications, based on: (a) use of low-powered hardware devices, (b) deploying mobile/relay agents for data collection, (c) clustering, and (d) data aggregation. Amongst these, data aggregation is widely acknowledged as an important tool to conserve energy in WSN and IoTs. The paper provides a comprehensive survey of various data aggregation strategies, discusses the efficacy of these strategies in handling issues that are typical to WSN and IoT applications. These issues severely impact the performance metrics such as: energy efficiency, latency, fault-tolerance, network throughput, and network lifetime. Therefore, to optimize the data aggregation approach, an application developer needs to arrive at optimal tradeoffs between these parameters. A major contribution of the paper is to present a holistic review of data aggregation approaches emphasizing the effect of topology, security, mobility, interference, and fault-tolerance in WSN and IoTs. Based on gap areas in literature, we throw open few challenges and present them as “posers”, and put-forth suggestions for further research.

Analysis of distributed ledger technologies for industrial manufacturing

In recent years, industrial manufacturing has undergone massive technological changes that embrace digitalization and automation towards the vision of intelligent manufacturing plants. With the aim of maximizing efficiency and profitability in production, an important goal is to enable flexible manufacturing, both, for the customer (desiring more individualized products) and for the manufacturer (to adjust to market demands). Manufacturing-as-a-service can support this through manufacturing plants that are used by different tenants who utilize the machines in the plant, which are offered by different providers. To enable such pay-per-use business models, Distributed Ledger Technology (DLT) is a viable option to establish decentralized trust and traceability. Thus, in this paper, we study potential DLT technologies for efficient and intelligent integration of DLT-based solutions in manufacturing environments. We propose a general framework to adapt DLT in manufacturing, and then we introduce the use case of shared manufacturing, which we utilize to study the communication and computation efficiency of selected DLTs in resource-constrained wireless IoT networks.

A Multimodel-Based Approach for Estimating Cause of Scanning Failure and Delay in IoT Wireless Network

The rapid growth in the IoT network comes with a huge security threat. Network scanning is considered necessary to identify vulnerable IoT devices connected to IP networks. However, most existing network scanning tools or system do not consider the burden of scan packet traffic on the network, especially in the IoT network where resources are limited. It is necessary to know the status of the communication environment and the reason why network scanning failed. Therefore, this paper proposes a multimodel-based approach which can be utilized to estimate the cause of failure/delay of network scanning over wireless networks where a scan packet or its response may sometimes be dropped or delayed. Specifically, the factors that cause network scanning failure/delay were identified and categorized. Then, using a machine learning algorithm, we introduced a multimodel linear discriminant analysis (MM-LDA) to estimate the cause of scan failure/delay based on the results of network scanning. In addition, a one-to-many model and a training data filtering technique were adopted to ensure that the estimation error was drastically reduced. The goal of our proposed method was to correctly estimate the causes of scan failure/delay in IP-connected devices. The performance of the proposed method was evaluated using computer simulation assuming a cellular (LTE) network as the targeted IoT wireless network and using LTE-connected devices as the targeted IoT devices. The proposed MM-LDA correctly estimates the cause of failure/delay of the network scan at an average probability of 98% in various scenarios. In comparison to other conventional machine learning classifiers, the proposed MM-LDA outperforms various classification methods in the estimation of the cause of scan failure/delay.

Process Optimization and Die Design of Hydromechanical Calendering Based on Wireless Network Communication and IoT

Aiming at the problem of the study, of high error probability in the practical application of traditional hydromechanical calendering process, the hydromechanical calendering process is optimized based on wireless network communication. GPRS or LORaipot-IOT wireless communication mode is adopted to calculate the impact kinetic energy of hydromechanical calendering, and the simulation model of hydromechanical calendering is built based on AMESim. The characteristics of hydromechanical calendering are analyzed. The oil pressure is controlled by setting the pressure adjusting valve and adjusting the spring pressure to realize the optimization of the hydromechanical calendering process. On this basis, the grinding tool is designed, and the clearance between punch and die is reserved to prevent the blank from cracking. The experimental results show that the error probability of hydromechanical calendering with the optimized design process is significantly lower than that of the control group, which can solve the problem of high error probability of traditional hydromechanical calendering process in practical application and meet the application requirements. Conclusions. Hydraulic mechanical calendering may be used to calendar different plastic materials, with a blank thickness of 0.5-3 mm, according to the technology’s operating principle, test production, and foreign data. Thicker sheets may be calendered with adequate hydraulic pressure. For the form of parts, the optimum effect is that the projectile, cone, etc. section vary from small too big. For the section does not change or smaller parts also have an impact? Some pieces with an interior concave shape and bottom and side stiffeners are difficult to the calendar. Hydromechanical calendering technologies make it easier to create these pieces.

A Study on Impact of DIS flooding Attack on RPL-based 6LowPAN Network

RPL is the default routing protocol for IPv6-based Low-Power Wireless Personal Area Networks (6LoWPAN). In an RPL-based 6LoWPAN, wireless sensor nodes or IoT devices connect to the network via DODAG Information Solicitation (DIS) packets with a predetermined time interval. A compromised or malicious node can use the abovementioned technique to broadcast unauthorised DIS messages to neighboring nodes to conduct routing attacks, such as DIS flooding attacks. In this paper, we conduct a simulation-based analysis of the impact of the DIS flooding attack on RPL-based 6LoWPAN under four distinct scenarios, each with two test cases. Scenario 1 is the reference topology, devoid of attacker nodes or test cases. From scenario 2 to scenario 4, the number of DIS attacker nodes increases from one to three to five. In each scenario, in test case A, all attacker nodes are deployed outside the sink node's radio range; in test case B, all attacker nodes are deployed within the sink node's radio range. Our study shows that increasing the number of attacker nodes and deployment location significantly negatively impacts PDR, E2ED, and power consumption. In the case of five DIS attackers nodes deployed within the sink node's radio range, the experimental result decreased PDR by 4.5%, E2ED increased by 62.98%, and the average power consumption increased by 126.7%.

A Counter-Eavesdropping Technique for Optimized Privacy of Wireless Industrial IoT Communications

The industrial Internet of Things (IIoTs) is a key component of the fourth industrial revolution (Industry 4.0) which is faced with privacy issues as the scale and sensitivity of user and system data constantly increases. Eavesdropping attack is one of such privacy issue of the IIoT system especially when the number of transmitting antennas is increased. Thus, the focus of this article is on establishing efficient privacy in an IIoT-multiple-input–multiple-output–multiple-antenna eavesdropping communications scenario. To achieve this, a closed-form derivation for asymptotic regularized prompt privacy rate is first formulated for IIoT network system. Then, the study further examines the design of optimal jamming parameters by proposing a model referred as optimal counter-eavesdropping channel approximation technique for tackling eavesdropping attack in IIoT. The simulated performance of the proposed model clearly shows that provided that the channel coherence time is less than two times the number of transmitting nodes, a high privacy precision is achieved even without deploying any artificial noise.

Cooperative Inference of DNNs for Delay- and Memory-Constrained Wireless IoT Systems

This work studies the cooperative inference of deep neural networks (DNNs), in which a memory-constrained end device performs a delay-constrained inference process with an aid of an edge server. Although several works considered the cooperative inference of DNNs in the literature, it was assumed in those works that the memory footprints at end devices are unlimited, which is in practice not realistic. To address this issue, in this work, a memory-aware cooperative DNN inference is proposed. Specifically, we propose to adopt knowledge distillation to obtain high-performing lightweight DNNs. To minimize the inference delay, we first analyze the end-to-end delay required for processing the proposed cooperative DNN inference, and then we minimize the delay by jointly optimizing the DNN partitioning point and the intermediate data transmission rate. Also, a dynamic DNN selection scheme is developed by fully exploiting the available memory resource in order to maximize the performance of the inference task in terms of inference accuracy. Experimental results demonstrate that the proposed cooperative DNN inference considerably outperforms the comparable schemes while satisfying both the delay constraint and the memory constraint.

Performance analysis of physical-layer network coding with QPSK modulation in wireless IoT networks

In this paper, we mathematically analyze bit-error rate (BER) and diversity order of physical-layer network coding (PNC) scheme in the two-way relay channel (TWRC) under the assumption of multiple antennas at relay node. We assume the quadrature phase shift keying (QPSK) modulation for simple data transmission in IoT applications, and max–min transmit antenna selection (TAS) scheme in the broadcast (BC) phase for achieving full diversity gain. In particular, we obtain the upper bound of BER at the multiple access (MA) phase, while deriving the exact BER performance at the BC phase. To the best our knowledge, the mathematical analysis in this paper is the first analytical results in the literature. Through computer simulations, the analytical results are validated.

Decentralized Wireless Federated Learning With Differential Privacy

This article studies decentralized federated learning algorithms in wireless IoT networks. The traditional parameter server architecture for federated learning faces some problems such as low fault tolerance, large communication overhead and inaccessibility of private data. To solve these problems, we propose a decentralized wireless federated learning algorithm called DWFL. The algorithm works in a system where the workers are organized in a peer-to-peer and server-less manner, and the workers exchange their privacy preserving data with the analog transmission scheme over wireless channels in parallel. With rigorous analysis, we show that DWFL satisfies <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$(\epsilon,\delta)$</tex-math></inline-formula> -differential privacy and the privacy budget per worker scales as <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$\mathcal {O}(\frac{1}{\sqrt{N}})$</tex-math></inline-formula> , in contrast with the constant budget in the orthogonal transmission approach. Furthermore, DWFL converges at the same rate of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">$\mathcal {O}(\sqrt{\frac{1}{TN}})$</tex-math></inline-formula> as the best known centralized algorithm with a central parameter server. Extensive experiments demonstrate that our algorithm DWFL also performs well in real settings.

A Secure and Efficient Method to Protect Communications and Energy Consumption in IoT Wireless Sensor Networks

The rapid growth of technology has resulted in the deployment of a large number of interconnected devices, resulting in a wide range of new societal services. Wireless sensor networks (WSNs) are a promising technology which is faced with the challenges of operating a large number of sensor nodes, information gathering, data transmission, and providing a means to act in different scenarios such as monitoring, surveillance, forest fire detection, and many others from the civil to military spectrum. The deployment scenario, the nature of the sensor-equipped nodes, and their communication methods make this architecture extremely vulnerable to attacks, tampering, and manipulation than conventional networks. Therefore, an optimal solution to ensure security in such networks which captures the major constraints of the network in terms of energy utilization, secured data transmission, bandwidth, and memory fingerprint to process data is required. This work proposes a fast, reliable, and secure method of key distribution and management that can be used to ensure the integrity of wireless sensor networks’ communications. Moreover, with regards to efficient energy utilization, an improvement of the Low Energy Adaptive Clustering Hierarchy (LEACH) algorithm (a cluster routing protocol that is mainly used in WSN) has been proposed to enhance the networks’ energy efficiency, simplicity, and load-balancing features. Therefore, in this paper, we propose a combination of a distributed key exchange and management methods based on elliptic curve cryptography to ensure security of node communication and an improved routing protocol based on the LEACH protocol to demonstrate better performance in parameters such as network lifespan, dead nodes, and energy consumption.

Digital Ecosystem Model for GIAHS: The Barroso Agro-Sylvo-Pastoral System

Globally Important Agricultural Heritage Systems (GIAHS) territories are highly relevant to achieving sustainable lifestyles with human subsistence in balance with the ecosystem. The Barroso agro-sylvo-pastoral system is a clear example of this alignment between existing society, nature and natural resources, the environment, landscapes, and contextual heritage. Moreover, the sustainable use of existing natural resources, excellent environmental conditions, and breath-taking untouched landscapes represent a truly relevant factor towards the development of a region economy that is still greatly influenced by an engraved cultural, patrimonial, and agricultural heritage. Given the GIAHS classification attributed to that territory, the need arises to guarantee the existing environmental conditions. This context will allow the maintenance of the GIAHS classification, ensuring the quality of life in the region and stimulating its socio-economic development and overall sustainability. The present article describes a proposal for a digital ecosystem model aimed at the Barroso GIAHS, composed of four main functional hubs that actively interact with each other: smart environment, smart government, smart economy, and smart people. Based on wireless sensor networks, IoT, artificial intelligence, data analytics, and other technological solutions, this solution will allow real-time control of the territory’s environmental conditions and develop more efficient and well-supported management and governance.

Application placement with shared monitoring points in multi-purpose IoT wireless sensor networks

The main function of a wireless sensor network (WSN) is to gather data from a certain region and transfer the data to a center or remote locations for further processing. The collected data can be of interest for many applications. Therefore, a physical WSN owned by a single provider can be utilized by many customer applications. Additionally, the data of a particular point or sub-region can satisfy the need of multiple applications. Hence, sensing the data only once in such cases is beneficial to reduce the energy consumption, network traffic and acceptance ratio of the applications. We call this as monitoring point based shared data approach. In this paper, we focus on the placement of applications each of which requires several points to be monitored in an area a WSN covers. We first propose such a monitoring point based shared data approach for WSNs that will serve multiple dynamic applications. We also propose two methods for application placement over a shared physical WSN: one greedy method and one genetic algorithm based method called GABAP. We did extensive simulation experiments to evaluate our algorithms. The results show the effectiveness of our methods in fast and close-to-optimum placement of applications over a single network.

Benchmarking of lightweight cryptographic algorithms for wireless IoT networks

Benchmarking of lightweight cryptographic algorithms for wireless IoT networks

High-temperature operatable triboelectric nanogenerator using microdome-patterned polyimide for self-powered sensors

Renewable energy has received attention all around the world to deal with environmental pollution and energy crisis. Triboelectric nanogenerators (TENGs) has emerged as a promising way to transition to green energy. However, typical TENGs exhibit inferior output, low heat-resistance, and low durability. Polyimide is one of the most mechanically strong and thermally stable thermoplastics, and it could be used to overcome the drawbacks of TENGs. Thus, polyimide has become a promising building material for robust TENGs. Herein, we have rationally designed and created an efficient polyimide tribo-surface with customizable non-close packed microdome arrays (md-PI) to assemble TENGs (md-TENG) for harvesting windmill energy and vibration from engine even with hot air and high humidity surrounding. The md-TENG fabricated with md-PI and flat aluminum can generate an exceptional output power of 1.42 W∙m−2. This is about 29 times higher than the output of a TENG using a normal flat PI. Additional FEM simulations are performed to analyze the relationship between the contact area and mechanical stresses on the md-PI. Moreover, the md-TENG possesses outstanding durability of over 16,000 cycles of contact-separation, and the device exhibits better outputs at elevated temperature. The newly developed md-TENG achieves excellent thermal stability and durability as well as superior electrical output, which are essential to expand the practical applications of TENG. To illustrate the strengths of md-TENG, we demonstrated a wireless communication self-powered IoT humidity sensor and a vibration TENG sensor at high ambient air temperature in the engine compartment of a car.