Resource-limited Applications Research Articles

Enhanced lightweight encryption algorithm based on chaotic systems

In order to improve security and efficiency, this study presents a novel lightweight encryption technique that makes use of chaotic systems. Our method creatively combines the new chaotic KLEIN_64 algorithm with the Keccak-256 hash function, offering a solid basis for producing initial values essential for causing chaotic maps during the encryption process. After a deep validation with rigorous NIST testing, our chaotic pseudo random generator, LAC, exhibits excellent reliability and cryptographic robustness. Furthermore, the complexity of the cryptographic round function is improved by incorporating a second chaotic pseudo random generator that combines chaotic LFSR and Skew Tent Maps, thereby fortifying security measures.Designed with resource-limited applications in mind, our approach ensures that the cryptosystem remains both lightweight and efficient, meeting the stringent constraints typical of such environments. The practical feasibility and performance of our approach are extensively evaluated through FPGA implementation on the Zybo 7Z010 platform. Our implementation achieves a remarkable throughput of 2.820 Gbps while maintaining optimal resource utilization and efficiency. Extensive experimental results confirm the superior security of our cryptosystem, with correlation tests, entropy measurement, and histogram analysis showcasing robustness against statistical attacks. Moreover, the cryptosystem shows little fluctuation in the Unified Average Changing Intensity (UACI) and Non-Linear Pixel Change Rate (NPCR), confirming its resistance to differential attacks. Overall, our technique advances lightweight cryptography by providing a robust and efficient solution to modern cybersecurity challenges. In particular, our approach is well-suited for applications with limited resources, ensuring that security is maintained without compromising on performance or efficiency, thus fulfilling the needs of modern, constrained environments.

Exploration of an Open Vocabulary Model on Semantic Segmentation for Street Scene Imagery

This study investigates the efficacy of an open vocabulary, multi-modal, foundation model for the semantic segmentation of images from complex urban street scenes. Unlike traditional models reliant on predefined category sets, Grounded SAM uses arbitrary textual inputs for category definition, offering enhanced flexibility and adaptability. The model’s performance was evaluated across single and multiple category tasks using the benchmark datasets Cityscapes, BDD100K, GTA5, and KITTI. The study focused on the impact of textual input refinement and the challenges of classifying visually similar categories. Results indicate strong performance in single-category segmentation but highlighted difficulties in multi-category scenarios, particularly with categories bearing close textual or visual resemblances. Adjustments in textual prompts significantly improved detection accuracy, though challenges persisted in distinguishing between visually similar objects such as buses and trains. Comparative analysis with state-of-the-art models revealed Grounded SAM’s competitive performance, particularly notable given its direct inference capability without extensive dataset-specific training. This feature is advantageous for resource-limited applications. The study concludes that while open vocabulary models such as Grounded SAM mark a significant advancement in semantic segmentation, further improvements in integrating image and text processing are essential for better performance in complex scenarios.

A Lightweight Channel and Time Attention Enhanced 1D CNN Model for Environmental Sound Classification

One dimension convolutional neural networks (1D CNN) that directly take raw waveforms as input has less competition than 2D CNN recognizing environmental sound. In order to overcome its disadvantages, we propose a novel lightweight 1D CNN structure by employing attention mechanism, which has significant improvement in both accuracy and computational complexity. Concretely, (1) two attention modules are constructed along channel and time dimension separately, and combined to give an intermediate feature map, which focus on key frequency band and semantically related time frame information. (2) Without increasing training overhead, snapshot ensemble is employed to further improve performance. Results from two benchmarking datasets (UrbanSound8k, ESC-10) demonstrated that: by employing attention mechanism, our model outperforms all of the previously reported 1D CNN approaches in accuracy with less parameters. Meanwhile with improved performance gain, the proposed model is superior than most of the existing spectral-based 2D CNN approaches and competitive with SOTA performance, while with orders of magnitude parameters fewer. Overall, it indicates our model is compact and has good potential in practical resource-limited applications, such as sound recognition on embedded platform.

Improved 2-round collision attack on IoT hash standard ASCON-HASH

Lightweight cryptography algorithms are a class of ciphers designed to protect data generated and transmitted by the Internet of Things. They typically have low requirements in terms of storage space and power consumption, and are well-suited for resource-limited application scenarios such as embedded systems, actuators, and sensors. The NIST-approved competition for lightweight cryptography aims to identify lightweight cryptographic algorithms that can serve as standards. Its objective is to enhance data security in various scenarios. Among the chosen standards for lightweight cryptography, ASCON has been selected. ASCON-HASH is a hash function within the ASCON family. This paper presents a detailed analysis of the differential characteristics of ASCON-HASH, utilizing the quadratic S-box. Additionally, we employ message modification techniques and ultimately demonstrate a non-practical collision attack on the 2-round ASCON-HASH, requiring a time complexity of 298 hash function calls.

Broadband and large-aperture metasurface edge encoders for incoherent infrared radiation.

The prevalence of computer vision systems necessitates hardware-based approaches to relieve the high computational demand of deep neural networks in resource-limited applications. One solution would be to off-load low-level image feature extraction, such as edge detection, from the digital network to the analog imaging system. To that end, this work demonstrates incoherent, broadband, low-noise optical edge detection of real-world scenes by combining the wavefront shaping of a 24-mm aperture metasurface with a refractive lens. An inverse design approach is used to optimize the metasurface for Laplacian-based edge detection across the 7.5- to 13.5-μm LWIR imaging band, allowing for facile integration with uncooled microbolometer-based LWIR imagers to encode edge information. A polarization multiplexed approach leveraging a birefringent metasurface is also demonstrated as a single-aperture implementation. This work could be applied to improve computer vision capabilities of resource-constrained systems by leveraging optical preprocessing to alleviate the computational requirements for high-accuracy image segmentation and classification.

SCGNet: Shifting and Cascaded Group Network

Many lightweight networks have been proposed for resource-limited applications, however, they cannot be efficiently applied to neural-network processing units (NPUs) due to the limited operations supported by the NPUs, and few works focus on efficient network design on the NPUs. The basic blocks of networks such as MobileNetV2 and RegNet use smaller convolution kernels with relatively small receptive fields, which are not conducive to capturing large-scale spatial information. To address this weakness, we propose Shifting and Cascaded Group (SCG) block, where we cascade group convolutions with larger kernels to exploit multi-scale information and propose shifting group convolution to communicate channel information between different groups. Besides, we carefully devise our architecture guided by some principles and finally build a very efficient network called Shifting and Cascaded Group Network (SCGNet) on NPUs. To verify the superiority of our method, we conduct extensive experiments on various tasks including image classification, object detection, human pose estimation, person re-identification, and semantic segmentation to comprehensively evaluate the performance. Results on widely used datasets such as ImageNet, PASCAL VOC, COCO, MPII, Market-1501, DukeMTMC-ReID, CUHK03, and Cityscapes demonstrate that the proposed network is a more effective network on the corresponding vision tasks.

AODV-EOCW: An Energy-Optimized Combined Weighting AODV Protocol for Mobile Ad Hoc Networks.

The Ad Hoc On-demand Distance Vector (AODV) is a routing protocol for mobile ad hoc networks (MANETs) and other wireless ad hoc networks. The vanilla AODV protocol is simple and easy to implement because it only uses the hop count as a routing metric. Single-metric route determination also causes problems, such as network congestion and energy exhaustion, which limit the usage of AODV in resource-limited applications. To solve these problems, the authors propose a new routing protocol that combines the analytic hierarchy process (AHP), the entropy weight method (EWM), and AODV. The proposed protocol uses energy, congestion, and the hop count as metrics and weights these three metrics using AHP and EWM. To address the importance of energy in applications, such as drones, the proposed protocol chooses different comparison matrices for AHP at different node residual energy levels. Finally, the node chooses the best route link according to the score (sum of weighted metrics). It is also suitable for wireless sensor networks because the proposed protocol considers the residual energy of the node. The simulation results show that the improved routing protocol can effectively reduce the average end-to-end delay and energy consumption and prolong the lifetime of the whole network.

A fully-mapped and energy-efficient FPGA accelerator for dual-function AI-based analysis of ECG.

In this paper, a fully-mapped field programmable gate array (FPGA) accelerator is proposed for artificial intelligence (AI)-based analysis of electrocardiogram (ECG). It consists of a fully-mapped 1-D convolutional neural network (CNN) and a fully-mapped heart rate estimator, which constitute a complementary dual-function analysis. The fully-mapped design projects each layer of the 1-D CNN to a hardware module on an Intel Cyclone V FPGA, and a virtual flatten layer is proposed to effectively bridge the feature extraction layers and fully-connected layer. Also, the fully-mapped design maximizes computational parallelism to accelerate CNN inference. For the fully-mapped heart rate estimator, it performs pipelined transformations, self-adaptive threshold calculation, and heartbeat count on the FPGA, without multiplexed usage of hardware resources. Furthermore, heart rate calculation is elaborately analyzed and optimized to remove division and acceleration, resulting in an efficient method suitable for hardware implementation. According to our experiments on 1-D CNN, the accelerator can achieve 43.08× and 8.38× speedup compared with the software implementations on ARM-Cortex A53 quad-core processor and Intel Core i7-8700 CPU, respectively. For the heart rate estimator, the hardware implementations are 25.48× and 1.55× faster than the software implementations on the two aforementioned platforms. Surprisingly, the accelerator achieves an energy efficiency of 63.48 GOPS/W, which obviously surpasses existing studies. Considering its power consumption is only 67.74mW, it may be more suitable for resource-limited applications, such as wearable and portable devices for ECG monitoring.

Machine Learning for Microcontroller-Class Hardware: A Review.

The advancements in machine learning opened a new opportunity to bring intelligence to the low-end Internet-of-Things nodes such as microcontrollers. Conventional machine learning deployment has high memory and compute footprint hindering their direct deployment on ultra resource-constrained microcontrollers. This paper highlights the unique requirements of enabling onboard machine learning for microcontroller class devices. Researchers use a specialized model development workflow for resource-limited applications to ensure the compute and latency budget is within the device limits while still maintaining the desired performance. We characterize a closed-loop widely applicable workflow of machine learning model development for microcontroller class devices and show that several classes of applications adopt a specific instance of it. We present both qualitative and numerical insights into different stages of model development by showcasing several use cases. Finally, we identify the open research challenges and unsolved questions demanding careful considerations moving forward.

Hierarchical and non-monotonic key-policy attribute-based encryption and its application

Hierarchical-authority attribute-based encryption (HABE) can resolve the performance bottleneck of the key generation center in the traditional attribute-based encryption (ABE) system. HABE has numerous applications, such as cloud computing and smart grid systems. Nevertheless, almost all of the existing HABE schemes are in the ciphertext-policy setting and do not support non-monotonic access structures. In this work, we propose the first key-policy hierarchical-authority ABE with non-monotonic access structures, dubbed KP-HABE-NMaCS, which supports the authority key delegation, constant size ciphertexts and the large universe simultaneously. We further define the security model of KP-HABE-NMaCS and prove the selective security of it in the standard model. Then we optimize our scheme to realize the outsourced decryption. Subsequently, we present the detailed theoretical analysis of our constructions and the existing HABE constructions in computation and storage costs, and meanwhile implement ours and conduct a series of experiments. Both results indicate that our KP-HABE-NMaCS construction makes improvements in expressiveness and features, and achieves efficiency comparable to the previous HABE constructions. Furthermore, they also show that our extension construction is suitable for resource-limited applications. Eventually, we explore how to apply our KP-HABE-NMaCS to build an ABE with equality test and time-based authorization.

Compact modular multiplier design for strong security capabilities in resource-limited Telehealth IoT devices

Telehealth is an emerging model of delivering quality health to remote communities and stay-at-home users. This is motivated by the rising health care costs and by the benefits of many patients staying at home as opposed to extended hospital stays. Telehealth relies on IoT technology, but IoT devices present the weakest security link to the system. The challenge is to implement strong security capabilities in resource-limited IoT devices. This justifies the use of elliptic curve cryptography (ECC) over the other traditional and resource-consumed approaches such as RSA. Efficient modular multiplication is a basic operation needed for ECC systems. Therefore, the compact and efficient implementation of this operation will significantly affect the performance of ECC in resource-limited applications. This work presents a compact serial-in/serial-out word-based systolic implementation of modular multiplication. The proposed structure is derived using a formal and systematic technique for mapping regular iterative algorithms (RIA) onto processor arrays. The proposed methodology enables control of the processor array workload as well as the workload of each processing element. Controlling the processor word size allows control of system speed, latency, and area. The proposed processor structure saves area and energy consumption by a factor up to 96.3% and 98.5%, respectively.

Joint Coding/Decoding Optimization for DC-BICM System: Collaborative Design

Differential chaotic bit-interleaved coded modulation (DC-BICM) system can achieve coding gain from channel code without loss of bandwidth in the differential chaos shift keying (DCSK) system, but the current DC-BICM system using the belief propagation decoding algorithm has high decoding complexity, which brings burdens to resource-limited applications. In this letter, a novel collaborative design framework based on joint coding and decoding optimization is proposed for the DC-BICM system over multipath Rayleigh channels. Firstly, the column-weight-based (CWB) shuffled scheduling decoding is introduced into the DC-BICM system to lower the decoding complexity, called shuffled-based DC-BICM system. Then, a CWB extrinsic information transfer (CWB-EXIT) algorithm is developed for the theoretical analysis of this system. With the CWB-EXIT algorithm, a low-storage protograph low-density parity-check (P-LDPC) code is designed for this system, and the column-weight distribution is taken as an important constraint in the collaborative design process. In this way, the storage space used for storing the scheduling order in the CWB shuffled scheduling can be saved, whose size is proportional to code length. Theoretical analysis and simulations show that the shuffled-based DC-BICM system with the low-storage P-LDPC code can achieve lower decoding complexity, less storage requirement and better performance than the existing systems.

A self-powered pump based on gas-dissolved-in-liquid phenomenon to generate both negative and positive driving pressures

A portable, self-powered multifunctional pump plays an important role in propelling microfluidic devices used in resource-limited applications, such as easily operated low-cost devices in field applications and simplified experiment setups in college laboratories. In particularly, a self-powered pumping device that can provide both positive and negative driving pressures with large driving force is highly desired. In this paper, we present a novel self-powered microfluidic pump based on the negative pressure generated by gas dissolution in liquid. Based on a pressure transformation mechanism, this pump can supply not only the negative pressure but also the positive one. The maximum pumping volume is around 15 mL and the flow rate can be regulated from around 0.1 mL/min to 0.6 mL/min by altering the flowing resistance. The pumping power is generated from the carbon dioxide dissolution in the sodium hydroxide solution. Then the generated negative pressure is linked to the pressure transformation mechanism which transfers the pressure into negative pressure or positive pressure. As a demonstration of its high driving pressure, the proposed self-powered pump has been successfully applied in the droplet generation. Such high driving capacity and flexibility of the presented self-powered pump shows its potentials to be a new type of passive pumps with wide applications in microfluidics.

Suitability of Hood Geometry for Design of a PCM Neonate Incubator for Resource-Limited Clinical Applications.

To reduce dependence on electric-powered incubators, a number of alternate heat sources have been proposed. Phase change materials (PCM) are one of such because of their availability and cost effectiveness in rural areas. This study intends to explore the use of phase change material (PCM) such as paraffin wax as an alternative heat source over a variety of incubator hood geometries. This study presents three incubator hood geometries and their respective effects on maximum hood temperatures and time to reach these temperatures for a mainstream incubator. The three designs, cubic, pyramidal, and oval, were created using CAD software; mathematical computations for heat transfer analysis were undertaken using COMSOL Multiphysics software. Results show the maximum temperatures reached in the hoods were 308, 314.5, and 315K for the cubic, pyramidal, and oval-shaped geometries respectively. This offers a promising application of PCM-based as a choice material for incubator design for rural applications.

Human Comfortability: Integrating Ergonomics and Muscular-Informed Metrics for Manipulability Analysis During Human-Robot Collaboration

The ability to compute a quality index for manipulation tasks, in different configurations, has been widely used in robotics. However, it is poorly explored in human manipulation and physical human-robot collaboration (pHRC). Existing works that evaluate efficiency of human manipulation often focus only on heurisitic-based, biomechanics or ergonomics methods/tasks. Complementarity between these performance features allows for a better evaluation and more general criteria, applicable across tasks. This letter addressess this gap by generating a new metric that combines offline pre-computation of biomechanics, ergonomics, muscle assessment and joint constraints, and reducing the online time complexity, enhancing the response query time. The proposed solution allow us to build a quality distribution in the human's workspace which can be quickly tailored to specific tasks and filtered for design purposes. This method simplifies human manipulability assessment for both general and task-specific applications and, in contrast to existing works, is suitable for real-time and/or resource-limited applications. Numerical evidence shows the proposed analysis greatly outperforms previous results in terms of computing time without compromising performance.

ChannelNets: Compact and Efficient Convolutional Neural Networks via Channel-Wise Convolutions.

Convolutional neural networks (CNNs) have shown great capability of solving various artificial intelligence tasks. However, the increasing model size has raised challenges in employing them in resource-limited applications. In this work, we propose to compress deep models by using channel-wise convolutions, which replace dense connections among feature maps with sparse ones in CNNs. Based on this novel operation, we build light-weight CNNs known as ChannelNets. ChannelNets use three instances of channel-wise convolutions; namely group channel-wise convolutions, depth-wise separable channel-wise convolutions, and the convolutional classification layer. Compared to prior CNNs designed for mobile devices, ChannelNets achieve a significant reduction in terms of the number of parameters and computational cost without loss in accuracy. Notably, our work represents an attempt to compress the fully-connected classification layer, which usually accounts for about 25 percent of total parameters in compact CNNs. Along this new direction, we investigate the behavior of our proposed convolutional classification layer and conduct detailed analysis. Based on our in-depth analysis, we further propose convolutional classification layers without weight-sharing. This new classification layer achieves a good trade-off between fully-connected classification layers and the convolutional classification layer. Experimental results on the ImageNet dataset demonstrate that ChannelNets achieve consistently better performance compared to prior methods.

Ephemeral Secrets: Multi-Party Secret Key Acquisition for Secure IEEE 802.11 Mobile Ad Hoc Communication

Mobile ad hoc networks consist of wireless nodes and can be established quickly with minimal configuration and cost, because, they do not require any infrastructure in advance. Civil and military applications are using them extensively in emergency and mission-oriented scenarios respectively as multi-party communication systems. Whereas, the multi-party secret key acquisition is one of the acute issues in these low resource wireless ad hoc networks, especially, which are based on IEEE 802.11ah and IEEE 802.11ba (low power WiFi), IEEE 802.15.4(Zigbee), BLE, IEEE 802.15.6 (body-worn or wearable) devices. In this study, a novel low cost and robust approach has been proposed and tested to establish an identical secret key in a multi-user mobile ad hoc environment. We believe, it is one of the groundbreaking contributions toward establishing a cost-effective secret key acquisition solution with respect to memory, computation, and bandwidth. We have used Bloom-Filters to cope with these resource limitations of such wireless setups. The proposed approach has been tested using IEEE 802.11 adapters in a real environment, and we found it to be highly suitable for wireless resource-limited applications.

Driver Pose Estimation Using Recurrent Lightweight Network and Virtual Data Augmented Transfer Learning

Driver poses recognition contains three tasks such as body joint, head angle, and face landmark estimation, which is of paramount interest for the advanced driver assistance systems (ADAS). Recently proposed methods intend to use deeper and more complicated networks to achieve better performance, which leads to heavy models that are not feasible for the resource limited applications such as ADAS. To resolve this issue, we have worked on the following aspects: 1) a lightweight network model, which is referred to as recurrent multi-task thin net (RM-ThinNet), has been proposed which was especially designed for the computationally and memory limited devices; 2) a recurrent structure has been introduced to handle the scale difference and dependency between different tasks, and this recurrence ensures the different tasks are accomplished at different stages and their outputs can augment each other; and 3) a virtual data synthesization pipeline and a couple transfer learning method have been presented, by which network can be learnt effectively by relatively a small number of real data. Comparisons with the state-of-the-art methods reveal the superiority of the proposed method and competitive performance can be achieved with smaller model size and faster speed.

One-Bit LFM Pulse Radar: Harmonic Analysis and Target Reconstruction

One-bit radar, adopting one-bit analog-to-digital converters to perform signal sampling and quantization, is a promising technology for low-cost and resource-limited applications. In this paper, effects of one-bit quantization on the linear signal processing for the one-bit linear frequency modulated pulse radar are studied and a target reconstruction approach is proposed. Firstly, characteristics of 3-order harmonics of the one-bit signal are analyzed. It is indicated that when conventional linear signal processing methods are carried out, 3-order harmonics may not only raise the detection background but also result in false alarms. Secondly, a two-stage approach jointly performing the linear and nonlinear signal processing is proposed to suppress harmonics and recover true targets. Specifically, in the first stage, the linear signal processing is carried out to suppress unmatched harmonics and dimension reduction. In the second stage, based on a dimension-reduced signal observation model, generalized approximate message passing is utilized to suppress matched harmonics and recover true targets. The effectiveness of the proposed approach is demonstrated by numerical results.

A streaming accelerator of Convolutional Neural Networks for resource-limited applications

A streaming accelerator of Convolutional Neural Networks for resource-limited applications