Resource-limited Devices Research Articles

Lightweight, Secure, Similar-Document Retrieval over Encrypted Data

Applications for document similarity detection are widespread in diverse communities, including institutions and corporations. However, currently available detection systems fail to take into account the private nature of material or documents that have been outsourced to remote servers. None of the existing solutions can be described as lightweight techniques that are compatible with lightweight client implementation, and this deficiency can limit the effectiveness of these systems. For instance, the discovery of similarity between two conferences or journals must maintain the privacy of the submitted papers in a lightweight manner to ensure that the security and application requirements for limited-resource devices are fulfilled. This paper considers the problem of lightweight similarity detection between document sets while preserving the privacy of the material. The proposed solution permits documents to be compared without disclosing the content to untrusted servers. The fingerprint set for each document is determined in an efficient manner, also developing an inverted index that uses the whole set of fingerprints. Before being uploaded to the untrusted server, this index is secured by the Paillier cryptosystem. This study develops a secure, yet efficient method for scalable encrypted document comparison. To evaluate the computational performance of this method, this paper carries out several comparative assessments against other major approaches.

Sustainable Circular Micro Index for Evaluating Virtual Substitution Using Machine Learning with the Path Planning Problem as a Case Study

Due to the problems resulting from the COVID-19 pandemic, for example, semiconductor supply shortages impacting the technology industry, micro-, small-, and medium-sized enterprises have been affected because the profitability of their business models depends on market stability. Therefore, it is essential to propose alternatives to mitigate the various consequences, such as the high costs. One attractive alternative is to replace the physical elements using resource-limited devices powered by machine learning. Novel features can improve the embedded devices’ (such as old smartphones) ability to perceive an environment and be incorporated in a circular model. However, it is essential to measure the impact of substituting the physical elements employing an approach of a sustainable circular economy. For this reason, this paper proposes a sustainable circular index to measure the impact of the substitution of a physical element by virtualization. The index is composed of five dimensions: economic, social, environmental, circular, and performance. In order to describe this index, a case study was employed to measure the path-planning generator for micro aerial vehicles developed using virtual simulation using machine-learning methods. The proposed index allows considering virtualization to extend the life cycle of devices with limited resources based on suggested criteria. Thus, a smartphone and the Jetson nano board were analyzed as replacements of specialized sensors in controlled environments.

Small-Scale Depthwise Separable Convolutional Neural Networks for Bacteria Classification

Bacterial recognition and classification play a vital role in diagnosing disease by determining the presence of large bacteria in the specimens and the symptoms. Artificial intelligence and computer vision widely applied in the medical domain enable improving accuracy and reducing the bacterial recognition and classification time, which aids in making clinical decisions and choosing the proper treatment. This paper aims to provide an approach of 33 bacteria strains’ automated classification from the Digital Images of Bacteria Species (DIBaS) dataset based on small-scale depthwise separable convolutional neural networks. Our five-layer architecture has significant advantages due to the compact model, low computational cost, and reliable recognition accuracy. The experimental results proved that the proposed design reached the highest accuracy of 96.28% with a total of 6600 images and can be executed on limited-resource devices of 3.23 million parameters and 40.02 million multiply–accumulate operations (MACs). The number of parameters in this architecture is seven times less than the smallest model listed in the literature.

Sparsity-control ternary weight networks

Deep neural networks (DNNs) have been widely and successfully applied to various applications, but they require large amounts of memory and computational power. This severely restricts their deployment on resource-limited devices. To address this issue, many efforts have been made on training low-bit weight DNNs. In this paper, we focus on training ternary weight {−1, 0, ＋1} networks which can avoid multiplications and dramatically reduce the memory and computation requirements. A ternary weight network can be considered as a sparser version of the binary weight counterpart by replacing some −1s or 1s in the binary weights with 0s, thus leading to more efficient inference but more memory cost. However, the existing approaches to train ternary weight networks cannot control the sparsity (i.e., percentage of 0s) of the ternary weights, which undermines the advantage of ternary weights. In this paper, we propose to our best knowledge the first sparsity-control approach (SCA) to train ternary weight networks, which is simply achieved by a weight discretization regularizer (WDR). SCA is different from all the existing regularizer-based approaches in that it can control the sparsity of the ternary weights through a controller α and does not rely on gradient estimators. We theoretically and empirically show that the sparsity of the trained ternary weights is positively related to α. SCA is extremely simple, easy-to-implement, and is shown to consistently outperform the state-of-the-art approaches significantly over several benchmark datasets and even matches the performances of the full-precision weight counterparts.

Software Design and Experimental Evaluation of a Reduced AES for IoT Applications

IoT devices include RFID tags, microprocessors, sensors, readers, and actuators. Their main characteristics are their limited resources and computing capabilities, which pose critical challenges to the reliability and security of their applications. Encryption is necessary for security when using these limited-resource devices, but conventional cryptographic algorithms are too heavyweight and resource-demanding to run on IoT infrastructures. This paper presents a lightweight version of AES (called LAES), which provides competitive results in terms of randomness levels and processing time, operating on GF(24). Detailed mathematical operations and proofs are presented concerning LAES rounds design fundamentals. The proposed LAES algorithm is evaluated based on its randomness, performance, and power consumption; it is then compared to other cryptographic algorithm variants, namely Present, Clefia, and AES. The design of the randomness and performance analysis is based on six measures developed with the help of the NIST test statistical suite of cryptographic applications. The performance and power consumption of LAES on a low-power, 8-bit microcontroller unit were evaluated using an Arduino Uno board. LAES was found to have competitive randomness levels, processing times, and power consumption compared to Present, Clefia, and AES.

IoT Notary : Attestable Sensor Data Capture in IoT Environments

Contemporary IoT environments, such as smart buildings, require end-users to trust data-capturing rules published by the systems. There are several reasons why such a trust is misplaced—IoT systems may violate the rules deliberately or IoT devices may transfer user data to a malicious third-party due to cyberattacks, leading to the loss of individuals’ privacy or service integrity. To address such concerns, we propose IoT Notary , a framework to ensure trust in IoT systems and applications. IoT Notary provides secure log sealing on live sensor data to produce a verifiable “proof-of-integrity,” based on which a verifier can attest that captured sensor data adhere to the published data-capturing rules. IoT Notary is an integral part of TIPPERS, a smart space system that has been deployed at the University of California, Irvine to provide various real-time location-based services on the campus. We present extensive experiments over real-time WiFi connectivity data to evaluate IoT Notary , and the results show that IoT Notary imposes nominal overheads. The secure logs only take 21% more storage, while users can verify their one day’s data in less than 2 s even using a resource-limited device.

Dynamical Conventional Neural Network Channel Pruning by Genetic Wavelet Channel Search for Image Classification.

Neural network pruning is critical to alleviating the high computational cost of deep neural networks on resource-limited devices. Conventional network pruning methods compress the network based on the hand-crafted rules with a pre-defined pruning ratio (PR), which fails to consider the variety of channels among different layers, thus, resulting in a sub-optimal pruned model. To alleviate this issue, this study proposes a genetic wavelet channel search (GWCS) based pruning framework, where the pruning process is modeled as a multi-stage genetic optimization procedure. Its main ideas are 2-fold: (1) it encodes all the channels of the pertained network and divide them into multiple searching spaces according to the different functional convolutional layers from concrete to abstract. (2) it develops a wavelet channel aggregation based fitness function to explore the most representative and discriminative channels at each layer and prune the network dynamically. In the experiments, the proposed GWCS is evaluated on CIFAR-10, CIFAR-100, and ImageNet datasets with two kinds of popular deep convolutional neural networks (CNNs) (ResNet and VGGNet). The results demonstrate that GNAS outperforms state-of-the-art pruning algorithms in both accuracy and compression rate. Notably, GNAS reduces more than 73.1% FLOPs by pruning ResNet-32 with even 0.79% accuracy improvement on CIFAR-100.

Maximizing the latency fairness in UAV‐assisted MEC system

Unmanned aerial vehicles (UAV) assisted edge computing has risen as an assuring technique to accommodate ubiquitous edge computation for resource-limited devices. Thus, this paper proposes an approach to maximize the latency fairness in a UAV-assisted multi-access edge computing (MEC) system. To maximize latency fairness, the authors focus on minimizing the maximum latency experienced among the users. In here, multiple ground users (GUs) offload their tasks to MEC UAV in the absence or unavailability of ground servers due to a disaster or heavy traffic where an iterative algorithm is proposed to minimize the maximum latency among the users subject to minimum control link rate and total power constraints. Sequentially, the UAVs' 3D location, offloading ratio, GUs' transmit power and GUs' computational capacity are optimized. The location of the UAV is optimized by using the novel approach, guided pattern search algorithm while the altitude of the UAV is optimized by analyzing the elevation angle dependant behaviour of the channel gain. A simple approach is utilized for optimizing the offloading ratio of the users by considering the problem as minimizing the point-wise maximum of two convex functions while the bisection method is used to optimize the power allocation. Numerical simulation results illustrate that the proposed approach outperforms other baseline approaches in convergence, minimizing the maximum latency and maximizing and maintaining the fairness among the GUs. Furthermore, it is proved that the guided pattern search algorithm converges at least 3.5 times better while the proposed combined optimization gives 400% fairness gain, in comparison with the baseline approach.

Efficient topology control of blockchain peer to peer network based on SDN paradigm

Blockchain allows to securely store, using cryptography functions, validated transactions and other data across its peer to peer (P2P) network. This P2P network is generated and maintained by the participating peers in a distributed manner through peer discovery, neighbor selection and managing inbound/outbound connections. As these tasks require extra consumption of network bandwidth, energy, memory and CPU usage, resource-limited devices may not support inherent blockchain applications. Further, changes in the topology control functions or reconfiguration should be considered by all peers to fully benefit from the new performance, which can take time. The aim of this paper is to build a blockchain network in a manner that greatly reduce topology control overhead while guaranteeing the properties such as high flexibility, fast reconfigurability, connectivity, small diameter and clustering. For this, we propose to use the Software-Defined Networking (SDN) paradigm to manage the blockchain P2P network. This way, the topology control tasks are moved off the peers to a secure overlay layer composed of multiple servers having synchronized databases. This new layer is responsible for building and managing the topological structure of the P2P network layer based on random r-out digraphs. Next, we mathematically discuss r-out digraphs generation using binomial distribution and preferential attachment models. Then, and in order to reduce the number of connections per peer, we establish lower and minimum upper bounds on outbound and inbound connections respectively that still guarantee the P2P network feasibility and connectivity. Further, we investigate topological properties of blockchain P2P network such as connectivity, diameter and clustering. Finally, we provide extensive simulation and numerical results to verify the efficiency of our approach and illustrate the effects of centralized topology control on network performance.

Emerging 5G IoT Smart System Based on Edge-to-Cloud Computing Platform

Security over internet communication has now become difficult as technology is increasingly more effective and faster, particularly in resource limited devices such as wireless sensors, embedded devices, internet of things (IoT), radio frequency identification (RFID) tags, etc. However, IoT is expected to connect billions of computers as a hopeful technology for the future. Hence, security, privacy, and authentication services must protect the communication in IoT. There are several recent considerations, such as restricted computing capacity, register width, RAM size, specific operating environment, ROM size, etc. that have compelled IoT to utilize conventional measures of security. These technologies require greater data speeds, high throughput, expanded power, lower bandwidth, and high efficiency. In addition, IoT has transformed the world in light of these new ideas by offering smooth communication between heterogeneous networks (HetNets).

Revisiting Binary Local Image Description for Resource Limited Devices

The advent of a panoply of resource limited devices opens up new challenges in the design of computer vision algorithms with a clear compromise between accuracy and computational requirements. In this letter we present new binary image descriptors that emerge from the application of triplet ranking loss, hard negative mining and anchor swapping to traditional features based on pixel differences and image gradients. These descriptors, BAD (Box Average Difference) and HashSIFT, establish new operating points in the state-of-the-art's accuracy vs. resources trade-off curve. In our experiments we evaluate the accuracy, execution time and energy consumption of the proposed descriptors. We show that BAD bears the fastest descriptor implementation in the literature while HashSIFT approaches in accuracy that of the top deep learning-based descriptors, being computationally more efficient. We have made the source code public <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sup> .

Guest Editorial: Aerial Computing: Drones for Multi-Access Edge Computing

The employment of multi-access edge computing (MEC) promotes the emergence of novel computation-intensive and time-sensitive applications at the network edge. Resource-limited devices can offload their computing tasks to edge servers, thereby avoiding heavy computing loads and reducing energy consumption. However, the computing servers are usually embedded in the fixed access points (APs) or base stations (BSs) in traditional MEC, which make it difficult to meet the unexpected explosively increasing computing demands and server users in certain emergency scenarios. Due to mobility, flexibility, and maneuverability, aerial computing has drawn extensive attention, where unmanned aerial vehicles (UAVs) equipped with computing servers provide agile computing services to mobile devices. Despite the potential of the aerial computing, however, many challenges also need to be addressed in this new paradigm. Until now, limited research progress has been made toward aerial computing.

Binary neural networks for image super-resolution

Recently, deep convolutional neural networks (DNNs) have achieved state-of-the-art performance on the super-resolution task. However, bigger and deeper networks often lead to high computational cost and high memory usage, preventing massive applications on resource-limited devices. Therefore, reducing the model storage and computation costs will expand the application range of DNNs. Binary neural networks, which could reduce the model size and allow for efficient inference, are energy-efficient for embedded devices. It motivates us to apply the binary methods into super-resolution field, enabling us to meet the requirements of the hardware platforms in practical applications. Therefore, this paper focuses on the binary image super-resolution network. To this end, we first review the current binary neural networks and introduce the technical details and algorithm characteristics of the existing algorithms in detail. Then, we propose to apply the existing binary training methods to the super-resolution task. On this basis, we propose a new binary super-resolution method, which advances network performance by improving the representation ability of the quantized network in the forward propagation and reducing information loss in the backward propagation. Extensive experiments demonstrate that the proposed method not only outperforms existing binary methods but also outperforms the state-of-the-arts binary super-resolution methods.

Improving Variational Autoencoder based Out-of-Distribution Detection for Embedded Real-time Applications

Uncertainties in machine learning are a significant roadblock for its application in safety-critical cyber-physical systems (CPS). One source of uncertainty arises from distribution shifts in the input data between training and test scenarios. Detecting such distribution shifts in real-time is an emerging approach to address the challenge. The high dimensional input space in CPS applications involving imaging adds extra difficulty to the task. Generative learning models are widely adopted for the task, namely out-of-distribution (OoD) detection. To improve the state-of-the-art, we studied existing proposals from both machine learning and CPS fields. In the latter, safety monitoring in real-time for autonomous driving agents has been a focus. Exploiting the spatiotemporal correlation of motion in videos, we can robustly detect hazardous motion around autonomous driving agents. Inspired by the latest advances in the Variational Autoencoder (VAE) theory and practice, we tapped into the prior knowledge in data to further boost OoD detection’s robustness. Comparison studies over nuScenes and Synthia data sets show our methods significantly improve detection capabilities of OoD factors unique to driving scenarios, 42% better than state-of-the-art approaches. Our model also generalized near-perfectly, 97% better than the state-of-the-art across the real-world and simulation driving data sets experimented. Finally, we customized one proposed method into a twin-encoder model that can be deployed to resource limited embedded devices for real-time OoD detection. Its execution time was reduced over four times in low-precision 8-bit integer inference, while detection capability is comparable to its corresponding floating-point model.

A Software-Defined-Networking-Enabled Approach for Edge-Cloud Computing in the Internet of Things

The proliferation of smart devices has led to a huge amount of data streaming in the Internet of Things (IoT). However, the resource-limited devices cannot satisfy the demands of computing-in-tensive but delay-sensitive applications. The data delivery among devices may be tampered with by malicious users. These pose new challenges to provide secure and intelligent services in IoT. Blockchain and reinforcement learning (RL) are promising techniques for establishing a secure environment and intelligent resource management. In this article, we introduce a novel software defined networking (SDN)-enabled architecture for edge-cloud orchestrated computing to support secure and intelligent services in IoT. We first introduce the SDN-enabled architecture by integrating cloud computing, edge computing, and IoT networks. Then we provide several applications of SDN-enabled architecture in edge-cloud orchestrated computing. Next, we propose the blockchain and RL envisioned solutions to implement secure and intelligent services in IoT. Moreover, a case study of blockchain- and RL-enabled secure and intelligent computing offloading is presented to validate its effectiveness. We finally provide our conclusion and discuss several promising research directions.

Modular Lightweight Network for Road Object Detection Using a Feature Fusion Approach

This article presents a modular lightweight network model for road objects detection, such as car, pedestrian, and cyclist, especially when they are far away from the camera and their sizes are small. Great advances have been made for the deep networks, but small objects detection is still a challenging task. In order to solve this problem, a majority of existing methods utilize complicated network or bigger image size, which generally leads to higher computation cost. The proposed network model is referred to as modular feature fusion detector (MFFD), using a fast and efficient network architecture for detecting small objects. The contribution lies in the following aspects: 1) two base modules have been designed for efficient computation: a) Front module reduces the information loss from raw input images and b) Tinier module decreases the model size and computation cost, while ensuring the detection accuracy; 2) by stacking the base modules, we design a context features fusion framework for multiscale object detection; and 3) the proposed method is efficient in terms of model size and computation cost, which is applicable for resource-limited devices, such as embedded systems for advanced driver-assistance systems (ADASs). Comparisons with the state-of-the-art on the challenging KITTI dataset reveal the superiority of the proposed method. Especially, 100 ft/s can be achieved on the embedded GPUs such as Jetson TX2.

A Lightweight Leakage-Resilient Identity-Based Mutual Authentication and Key Exchange Protocol for Resource-limited Devices

Identity (ID)-based mutual authentication and key exchange (IDMAKE) protocol for client-server environments is a critical cryptographic primitive. Recently, several leakage-resilient IDMAKE (LR-IDMAKE) protocols have been proposed to withstand a new type of attacks, named side-channel attacks. However, the existing LR-IDMAKE protocols have several drawbacks, especially, the performance problem for low-power clients. In this paper, a lightweight LR-IDMAKE protocol well-suited for resource-limited devices is proposed. As compared with the previously proposed LR-IDMAKE protocols, our protocol has the following merits: (1) It is the first lightweight LR-IDMAKE protocol without on-line pairing operation for clients; (2) During life cycle of the proposed protocol, it is resilient to continuous key leakage and has totally unbounded leakage property; (3) Security analysis is formally made to demonstrate that it is strongly secure against adversaries in the continuous-leakage extended-Canetti–Krawczyk (CLeCK) model. Eventually, performance experiences on an IoT device with low-computing power, i.e., the Raspberry PI, are conducted to demonstrate that our protocol is well-suited for resource-limited devices.

RISC-V Virtual Platform-Based Convolutional Neural Network Accelerator Implemented in SystemC

The optimization for hardware processor and system for performing deep learning operations such as Convolutional Neural Networks (CNN) in resource limited embedded devices are recent active research area. In order to perform an optimized deep neural network model using the limited computational unit and memory of an embedded device, it is necessary to quickly apply various configurations of hardware modules to various deep neural network models and find the optimal combination. The Electronic System Level (ESL) Simulator based on SystemC is very useful for rapid hardware modeling and verification. In this paper, we designed and implemented a Deep Learning Accelerator (DLA) that performs Deep Neural Network (DNN) operation based on the RISC-V Virtual Platform implemented in SystemC in order to enable rapid and diverse analysis of deep learning operations in an embedded device based on the RISC-V processor, which is a recently emerging embedded processor. The developed RISC-V based DLA prototype can analyze the hardware requirements according to the CNN data set through the configuration of the CNN DLA architecture, and it is possible to run RISC-V compiled software on the platform, can perform a real neural network model like Darknet. We performed the Darknet CNN model on the developed DLA prototype, and confirmed that computational overhead and inference errors can be analyzed with the DLA prototype developed by analyzing the DLA architecture for various data sets.

Adaptive time-bound access control for internet of things in fog computing architecture

This paper proposes an adaptive Time-Bound Attribute-Based Encryption Scheme (TB-ABE). TB-ABE is a fine-grained data access control scheme that maintains a secure data exchange environment between IoT devices. The lifetime of the system is divided into equal periods of time slots with unique encryption keys. TB-ABE combines Cyphertext-Policy Attribute Based Encryption with Time-bound keys. It only permits data users with attributes that satisfy the access policy set by the data owner to access the data for a specific period. It also proposes an efficient indirect revocation handling mechanism with minimal computations. The proposed scheme is implemented in Fog-Cloud computing architecture, where fog devices are used for partial outsourcing of encryption and decryption operations. This paper provides security and performance analysis for the proposed scheme in terms of computation, storage, communication and energy overheads. Analysis of proposed scheme proves its feasibility for limited-resource devices, while maintaining a secure communication between IoT devices.

Resource Management and Pricing for Cloud Computing Based Mobile Blockchain With Pooling

In a public blockchain system applying Proof of Work (PoW), the participants need to compete with their computing resources for reward, which is challenging for resource-limited devices. Mobile blockchain is proposed to facilitate the application of blockchain for mobile service, in which the lightweight devices can participate mining by renting resources from the Cloud Computing Service Provider (CCSP), but CCSP usually does not have the information about the demand preference of users. In this paper, a contract model is adopted to address the cloud computing resource allocation and pricing problem in the mobile blockchain. In particular, an adverse selection contract solution is proposed to overcome the information asymmetry problem, and resource pooling is introduced to improve the stability of users rewards. Simulation results show that the information asymmetry problem is well overcome by adverse selection contract so that CCSP can obtain more utility than linear pricing contracts. Furthermore, the resource pooling could effectively improve the users and CCSPs utilities. When the size of the mining pool is large enough, it can achieve an improvement effect of more than 10 times. The effect of pool size and user type distribution on CCSPs utility is also studied.