Lightweight Mobile Devices Research Articles

Hyperbolic Binary Neural Network.

Binary neural network (BNN) converts full-precision weights and activations into their extreme 1-bit counterparts, making it particularly suitable for deployment on lightweight mobile devices. While BNNs are typically formulated as a constrained optimization problem and optimized in the binarized space, general neural networks are formulated as an unconstrained optimization problem and optimized in the continuous space. This article introduces the hyperbolic BNN (HBNN) by leveraging the framework of hyperbolic geometry to optimize the constrained problem. Specifically, we transform the constrained problem in hyperbolic space into an unconstrained one in Euclidean space using the Riemannian exponential map. On the other hand, we also propose the exponential parametrization cluster (EPC) method, which, compared with the Riemannian exponential map, shrinks the segment domain based on a diffeomorphism. This approach increases the probability of weight flips, thereby maximizing the information gain in BNNs. Experimental results on CIFAR10, CIFAR100, and ImageNet classification datasets with VGGsmall, ResNet18, and ResNet34 models illustrate the superior performance of our HBNN over state-of-the-art methods.

(Invited, Digital Presentation) Copper Nanowire 3-D Interconnects for Mm-Wave Applications

Small, lightweight mobile devices are the beneficiaries of fabrication miniaturization that follow Moore’s law predictions, advances in complex integration concepts, and wireless technology. Successful implementations have therefore stimulated unprecedented growth in mobile applications. Within these miniaturized devices, contact electrodes with small form factors in some applications are needed but unfortunately present many challenges and can lead to unwanted parasitic effects. Identifying ways to meet this need without introducing parasitic effects is essential to minimize signal distortion and loss in integrated circuits. The plethora of applications, however, produce large number of users and also generate extremely large amounts of data that require higher data rates. Higher frequency designs satisfy bandwidth requirements and maintain miniature size; unfortunately, this comes at a cost. When operating frequencies increase, available power decreases due to transistor technology limitations and the corresponding shrinkage in circuit size results in higher resistance leading to more power consumption. To address this conundrum, approaches are needed that allow losses to much be managed, and ideally reduced to achieve the desired performance requirements.This talk will present the investigation of copper nanowire (CuNW) technology for use in millimeter wave regime (30 GHz to 300 GHz) interconnect design in silicon integrated circuits. The parallel nature of nanowires to behave like bundles offers the potential to minimize loss and parasitic effects at millimeter wave frequencies. Co-integration of CuNW into coplanar waveguide (CPW) interconnect configurations that are single and multi-layer architectures will be discussed. Two types of interconnects are studied: vertical via and CuNW integrated into CPW lines. The findings from the study of these structures will include a discussion on the fabrication and characterization methods used for each technology followed by the loss performance results observed for the different types of designs.At 40 and 60 GHz, vertical CuNW via loss values of 0.095 dB and 0.149 dB was observed, respectively. These responses were lower than those of other via technologies and comparable to those in anodized aluminum substrates. Thus, they were favorable and indicate that CuNW can be used in a co-integrate manner to incorporate small contact features into designs that can maintain or reduce loss in the millimeter wave regime. CuNWs integrated CPW interconnects can also be used as a transmission line. In this case, loss responses, observed in the millimeter wave regime, can be reduced. In one design, loss values of 0.48 dB/mm and 0.59 dB/mm at 90 and 180 GHz, respectively, were determined, suggesting that conductor loss can be reduced in upper millimeter regime. At these frequencies, higher order mode suppression is essential to minimize unwanted substrate losses. Another design, however, improved losses in the low RF and microwave frequency regime, suggesting that low frequency designs can also experience conductor loss reduction. The conclusion thus far is that CuNW co-integration into CPW lines show promise for improving the performance of interconnect design across the RF to millimeter-wave spectrum when designed carefully.

YOLOv8-CGRNet: A Lightweight Object Detection Network Leveraging Context Guidance and Deep Residual Learning

The growing need for effective object detection models on mobile devices makes it essential to design models that are both accurate and have fewer parameters. In this paper, we introduce a YOLOv8 Res2Net Extended Network (YOLOv8-CGRNet) approach that achieves enhanced precision under standards suitable for lightweight mobile devices. Firstly, we merge YOLOv8 with the Context GuidedNet (CGNet) and Residual Network with multiple branches (Res2Net) structures, augmenting the model’s ability to learn deep Res2Net features without adding to its complexity or computational demands. CGNet effectively captures local features and contextual surroundings, utilizing spatial dependencies and context information to improve accuracy. By reducing the number of parameters and saving on memory usage, it adheres to a ‘deep yet slim’ principle, lessening channel numbers between stages. Secondly, we explore an improved pyramid network (FPN) combination and employ the Stage Partial Spatial Pyramid Pooling Fast (SimPPFCSPC) structure to further strengthen the network’s capability in processing the FPN. Using a dynamic non-monotonic focusing mechanism (FM) gradient gain distribution strategy based on Wise-IoU (WIoU) in an anchor-free context, this method effectively manages low-quality examples. It enhances the overall performance of the detector. Thirdly, we introduce Unifying Object Detection Heads with Attention, adapting to various input scenarios and increasing the model’s flexibility. Experimental datasets include the commonly used detection datasets: VOC2007, VOC2012, and VisDrone. The experimental results demonstrate a 4.3% improvement in detection performance by the proposed framework, affirming superior performance over the original YOLOv8 model in terms of accuracy and robustness and providing insights for future practical applications.

Coriolis: enabling metagenomic classification on lightweight mobile devices.

The introduction of portable DNA sequencers such as the Oxford Nanopore Technologies MinION has enabled real-time and in the field DNA sequencing. However, in the field sequencing is actionable only when coupled with in the field DNA classification. This poses new challenges for metagenomic software since mobile deployments are typically in remote locations with limited network connectivity and without access to capable computing devices. We propose new strategies to enable in the field metagenomic classification on mobile devices. We first introduce a programming model for expressing metagenomic classifiers that decomposes the classification process into well-defined and manageable abstractions. The model simplifies resource management in mobile setups and enables rapid prototyping of classification algorithms. Next, we introduce the compact string B-tree, a practical data structure for indexing text in external storage, and we demonstrate its viability as a strategy to deploy massive DNA databases on memory-constrained devices. Finally, we combine both solutions into Coriolis, a metagenomic classifier designed specifically to operate on lightweight mobile devices. Through experiments with actual MinION metagenomic reads and a portable supercomputer-on-a-chip, we show that compared with the state-of-the-art solutions Coriolis offers higher throughput and lower resource consumption without sacrificing quality of classification. Source code and test data are available from http://score-group.org/?id=smarten.

Provably Secure Pseudo-Identity Three-Factor Authentication Protocol Based on Extended Chaotic-Maps for Lightweight Mobile Devices

Authentication and key agreement (AKA) play a major role in an open network environment to communicate between two or more participants securely. Authentication and key agreement protocols should protect sensitive data from a malevolent adversary by offering a several of services, such as user credential privacy and authentication when a user’s or server’s private key is revealed or a smart card is lost/stolen. Unfortunately, the majority of offered authentication and key agreement protocols suggested in the literature are vulnerable to many attacks. In this research, we present a pseudo-identity-based secure mutual authentication protocol that protects privacy. The suggested protocol resists replay attacks, de-synchronization attacks, eavesdropping attacks, denial of service attacks, insider attacks, known session-specific temporary information attacks, and man-in-the-middle attacks efficiently and also provides forward security. In order to offer privacy and prevent traceability, our protocol employs the pseudo-identity mechanism. We present a comprehensive security analysis, which includes formal verification with the Scyther tool; a BAN logic proof; and an informal discussion of various attacks, demonstrating that the proposed protocol is secure against all well-known attacks. Additionally, we give a performance study and a comparison to similar existing protocols.

KecNet: A Light Neural Network for Arrhythmia Classification Based on Knowledge Reinforcement.

Acquiring electrocardiographic (ECG) signals and performing arrhythmia classification in mobile device scenarios have the advantages of short response time, almost no network bandwidth consumption, and human resource savings. In recent years, deep neural networks have become a popular method to efficiently and accurately simulate nonlinear patterns of ECG data in a data-driven manner but require more resources. Therefore, it is crucial to design deep learning (DL) algorithms that are more suitable for resource-constrained mobile devices. In this paper, KecNet, a lightweight neural network construction scheme based on domain knowledge, is proposed to model ECG data by effectively leveraging signal analysis and medical knowledge. To evaluate the performance of KecNet, we use the Association for the Advancement of Medical Instrumentation (AAMI) protocol and the MIT-BIH arrhythmia database to classify five arrhythmia categories. The result shows that the ACC, SEN, and PRE achieve 99.31%, 99.45%, and 98.78%, respectively. In addition, it also possesses high robustness to noisy environments, low memory usage, and physical interpretability advantages. Benefiting from these advantages, KecNet can be applied in practice, especially wearable and lightweight mobile devices for arrhythmia classification.

Deep learning for site safety: Real-time detection of personal protective equipment

The leading causes of construction fatalities include traumatic brain injuries (resulted from fall and electrocution) and collisions (resulted from struck by objects). As a preventive step, the U.S. Occupational Safety and Health Administration (OSHA) requires that contractors enforce and monitor appropriate usage of personal protective equipment (PPE) of workers (e.g., hard hat and vest) at all times. This paper presents three deep learning (DL) models built on You-Only-Look-Once (YOLO) architecture to verify PPE compliance of workers; i.e., if a worker is wearing hard hat, vest, or both, from image/video in real-time. In the first approach, the algorithm detects workers, hats, and vests and then, a machine learning model (e.g., neural network and decision tree) verifies if each detected worker is properly wearing hat or vest. In the second approach, the algorithm simultaneously detects individual workers and verifies PPE compliance with a single convolutional neural network (CNN) framework. In the third approach, the algorithm first detects only the workers in the input image which are then cropped and classified by CNN-based classifiers (i.e., VGG-16, ResNet-50, and Xception) according to the presence of PPE attire. All models are trained on an in-house image dataset that is created using crowd-sourcing and web-mining. The dataset, named Pictor-v3, contains ~1,500 annotated images and ~4,700 instances of workers wearing various combinations of PPE components. It is found that the second approach achieves the best performance, i.e., 72.3% mean average precision (mAP), in real-world settings, and can process 11 frames per second (FPS) on a laptop computer which makes it suitable for real-time detection, as well as a good candidate for running on light-weight mobile devices. The closest alternative in terms of performance (67.93% mAP) is the third approach where VGG-16, ResNet-50, and Xception classifiers are assembled in a Bayesian framework. However, the first approach is the fastest among all and can process 13 FPS with 63.1% mAP. The crowed-sourced Pictor-v3 dataset and all trained models are publicly available to support the design and testing of other innovative applications for monitoring safety compliance, and advancing future research in automation in construction.

Discrimination of Sleep Spindles in Ear-EEG.

Sleep spindles are brief oscillatory events observed in EEG measurements during sleep, related to both sleep staging and basic neuroscience. The objective of this study was to investigate to which extent sleep spindles are observable from ear-EEG. The analysis was based on single-night recordings from 12 subjects, wearing both a polysomnography setup and two light-weight mobile EEG devices (ear-EEG). By introducing a sleep spindle index capable of discriminating between epochs with distinct spindles and distinctly spindle-free epochs, we describe to which extent the most clear cut sleep spindles (as labeled using scalp EEG) can be detected using ear-EEG. We find that ear-EEG can be used to detect sleep spindles, at a performance level similar to scalp derivations. We speculate that part of the observed discrepancy between ear-EEG and the gold standard (scalp EEG) could be caused by the visibility of different spindles in the ear-EEG.

Light-Weight and Privacy-Preserving Authentication Protocol for Mobile Payments in the Context of IoT

The widespread use of smart devices attracts much attention on the research for a mobile payment protocol in the context of the Internet of Things (IoT). However, payment trust and user privacy still raise critical concerns to the application of mobile payments since existing authentication protocols for mobile payments either suffer from the heavy workload on a resource-limited smart device or cannot provide user anonymity in the mobile payment. To address these challenges elegantly, this paper presents a lightweight and privacy-preserving authentication protocol for mobile payment in the context of IoT. First, we put forward a unidirectional certificateless proxy re-signature scheme, which is of independent interest. Based on this signature scheme, this paper, then, gives a new mobile payment protocol that for the first time not only achieves anonymity and unforgeability but also leaves low resource consumption on smart devices. In the proposed protocol, the efficiency is notably improved by placing the most computational cost on Pay Platform (usually with abundant computational power) instead of lightweight mobile devices. Moreover, by considering that the Pay Platform and Merchant Server needs to perform computation for each transaction, the idea of batch-verification has been adopted to mitigate the overhead for millions of users at the Pay Platform and Merchant Server to address the scalability issue. Through the formal security analysis presented in this paper, the proposed protocol is proved to be secure under the extended CDH problem. In addition, the performance evaluation shows that the proposed protocol is feasible and efficient for the resource-limited smart devices in the IoT.

Compact feature subset-based multi-label music categorization for mobile devices

Music categorization based on acoustic features extracted from music clips and user-defined tags forms the basis of recent music recommendation applications, because relevant tags can be automatically assigned based on the feature values and their relation to tags. In practice, especially for handheld lightweight mobile devices, there is a certain limitation on the computational capacity, owing to consumers’ usage behavior or battery consumption. This also limits the maximum number of acoustic features to be extracted, and results in the necessity of identifying a compact feature subset that is used for the music categorization process. In this study, we propose an approach to compact feature subset-based multi-label music categorization for mobile music recommendation services. Experimental results using various multi-labeled music datasets reveal that the proposed approach yields better performance when compared to conventional approach.

Collaborative Mobile Edge Computation Offloading for IoT over Fiber-Wireless Networks

Mobile edge computing is envisioned to be a promising paradigm to address the conflict between computationally intensive IoT applications and resource-constrained lightweight mobile devices. However, most existing research on mobile edge computation offloading has only taken the resource allocation between the mobile devices and the MEC servers into consideration, ignoring the huge computation resources in the centralized cloud computing center. To make full use of the centralized cloud and distributed MEC resources, designing a collaborative computation offloading mechanism becomes particularly important. Note that current MEC hosted networks, which mostly adopt the networking technology integrating cellular and core networks, face new challenges of single networking mode, long latency, poor reliability, high congestion, and high energy consumption. Hybrid fiber-wireless networks integrating both low-latency fiber optic and flexible wireless technologies should be a promising solution. Toward this end, we provide in this article a generic fiber-wireless architecture with coexistence of centralized cloud and distributed MEC for IoT connectivity. The problem of cloud-MEC collaborative computation offloading is defined, and a game-theoretic collaborative computation offloading scheme is proposed as our solution. Numerical results corroborate that our proposed scheme can achieve high energy efficiency and scales well as the number of mobile devices increases.

Ecological foraging models as inspiration for optimized recycling systems in the circular economy

Converting the consumer electronic product system from a linear system to a circular one has a number of key challenges. A mismatch is observed between the rapidly changing devices entering the market and the slowly evolving voluntary design policies, regulations, and e-waste processing business strategies. Conventional electronic waste (e-waste) management systems were historically optimized to extract high-value components from large products that were relatively easy to disassemble, but the products now entering the waste stream are more often light-weight mobile devices that are typically not covered by regulations or do not contain as high a concentration of valuable metals. This article proposes that transformations in the e-waste processing system aimed at closing the material loop should look to the circular processes found in natural ecosystems, which have evolved to optimize closed loop nutrient cycling. Like species in nature, e-waste processors make decisions about where and what to “eat,” balancing a food’s quality and abundance with the energy expended in obtaining it. Adapting the concept of optimal foraging theory, we demonstrate here a conceptual framework that draws parallels between foraging behavior in the ecological and industrial world, evaluates four potential mathematical models that case be applied to the e-waste case, and demonstrate how optimal foraging decisions can guide business, design, and end of life management toward circular economy goals in the consumer electronic system.

Transparent Computing

The guest editor highlights the topics touched on in the articles in this special issue. As we enter the Internet-of-Things era, in which lightweight mobile devices become the main online terminals in daily life, transparent computing promises great opportunities as well as new challenges.

Efficient Attribute-Based Comparable Data Access Control

With the proliferation of mobile devices in recent years, there is a growing concern regarding secure data storage, secure computation, and fine-grained access control in data sharing for these resource-constrained devices in a cloud computing environment. In this work, we propose a new efficient framework named Constant-size Ciphertext Policy Comparative Attribute-Based Encryption (CCP-CABE) with the support of negative attributes and wildcards. It embeds the comparable attribute ranges of all the attributes into the user’s key, and incorporates the attribute constraints of all the attributes into one piece of ciphertext during the encryption process to enforce flexible access control policies with various range relationships. Accordingly, CCP-CABE achieves the efficiency because it generates constant-size keys and ciphertext regardless of the number of involved attributes, and it also keeps the computation cost constant on lightweight mobile devices. We further discuss how to extend CCP-CABE to fit a scenario with multiple attribute domains, such that the decryption proceeds from the least privileged attribute domain to the most privileged one to help protect the privacy of the access policy. We provide security analysis and performance evaluation to demonstrate their efficiency at the end.

A Low Complexity System Based on Multiple Weighted Decision Trees for Indoor Localization

Indoor position estimation has become an attractive research topic due to growing interest in location-aware services. Nevertheless, satisfying solutions have not been found with the considerations of both accuracy and system complexity. From the perspective of lightweight mobile devices, they are extremely important characteristics, because both the processor power and energy availability are limited. Hence, an indoor localization system with high computational complexity can cause complete battery drain within a few hours. In our research, we use a data mining technique named boosting to develop a localization system based on multiple weighted decision trees to predict the device location, since it has high accuracy and low computational complexity. The localization system is built using a dataset from sensor fusion, which combines the strength of radio signals from different wireless local area network access points and device orientation information from a digital compass built-in mobile device, so that extra sensors are unnecessary. Experimental results indicate that the proposed system leads to substantial improvements on computational complexity over the widely-used traditional fingerprinting methods, and it has a better accuracy than they have.

A road network modeling method for map matching on lightweight mobile devices

With proliferation of lightweight mobile devices such as mobile phones and explosion of location-based social networking services, there is a growing demand for matching between geographic locations and road networks on mobile devices. Nonetheless, existing methods for map matching only focus on accuracy and/or efficiency improvement, whereas they seldom take into account the capacity for storing map data and energy consumption during the matching process. This paper presents a method that is specifically designed for lightweight mobile devices with limited storage and computing resources, thereby providing an effective solution for map matching on mobile and embedded environments. Extensive experiments were carried out to compare proposed method against traditional approaches. The results indicate that our method can cut down the storage cost for road networks by 75 % compared to traditional methods, with only 3---5 % extra running time, which demonstrates the practical usefulness and superiority of our proposal in real-world mobile applications.

Context-aware computing

A 2011 Gartner report [3] describes context-aware computing as a game-changing opportunity for enterprises to improve both productivity and profits. Context-aware computing is about making applications and content more relevant to a user's context, e.g., when and where the user is, thereby improving user experience. For instance, a coupon delivered to a user at a wrong time or at a wrong location is considered a nuisance. On the contrary, receiving a timely, usable coupon before purchasing a merchandise is a treat. Context-aware computing is not a new concept, but the ongoing mobile revolution makes it both necessary and feasible. • Necessary because the mobile phone display is small and information must be delivered with much higher relevance and precision to meet user needs. • Feasible because small, light-weight mobile devices allow users to almost always carry them around, and much can be learned via a phone about its user's habits and states. Context-aware computing involves first acquiring context and then taking context-dependent actions. For instance, a phone can sense a user's location and turn off its GPS unit to conserve power when the user enters a building, or it can collect EKG signals of a user and trigger an alert if the user's heart beats irregularly. Similarly, a restaurant can send a coupon to a user when that user is queued up in front of a nearby restaurant. The useful context can be divided into three categories: information on the user (knowledge of habits, emotional state, biophysiological conditions), the user's environment (time, location, co-location of others, social interaction), and the user's tasks (transportation mode, engaged tasks, general goals) [4]. Context-aware computing can be applied to benefit applications in many areas including but not limited to information retrieval, facility management, productivity enhancement, in addition to the aforementioned three examples representing power management, health care, and commerce, respectively.

A Method for Estimating Angular Separation in Mobile Wireless Sensor Networks

Resource-constrained mobile sensors require periodic position measurements for navigation around the sensing region. Such information is often obtained using GPS or onboard sensors such as optical encoders. However, GPS is not reliable in all environments, and odometry accrues error over time. Although several localization techniques exist for wireless sensor networks, they are typically time consuming, resource intensive, and/or require expensive hardware, all of which are undesirable for lightweight mobile devices. In this paper, we describe a technique for determining spatial relationships that is suitable for resource-constrained mobile sensors. Angular separation between multiple pairs of stationary sensor nodes is derived using wheel encoder data in conjunction with the measured Doppler shift of an RF interference signal. Our experimental results demonstrate that using this technique, a robot is able to determine the angular separation between four pairs of sensors in a 45 × 35 m sensing region with an average error of 0.28 rad. in 0.68 s.

Exploiting social interactions using opportunistic networks

Social interaction has drastically evolved over time. Moving away from face-to-face based interactions, telephone networks made the first step towards remote social interaction. The internet, further enhanced with the tremendous increase in lightweight mobile devices, has taken social interaction to new frontiers. Users can already email, chat, call, and video conference with others from around the world without necessarily being attached to any fixed location. The final frontier has been to exploit this technology to completely virtualize social interaction via online social networking services such as facebook, orkut, MySpace, or LinkedIn, etc. These applications create a virtual world where users build social networks of their acquaintances and allow people belonging to these social networks or communities to freely interact regardless of the boundaries of time and location. At this point, we pose a simple question, is this truly the final frontier with respect to social interaction? When people with similar interests or common acquaintances are a short distance from one another, like the same street, train, or mall, these people have no mechanism to identify this potential social interaction. Current research in geolocalization applications running on mobile devices provide some solutions to such problems, however, they face numerous challenges including network coverage, cost, and energy consumption concerns. We ultimately need context aware, adaptive, and agile solutions that can seamlessly extend peoples senses beyond their physical boundaries in order to exploit potentially rewarding social interactions. Our work, targeted towards fulfilling this need, takes advantage of physical context merged with online social relationships to ultimately improve the physical social interaction experience of people. We will discuss how current research thrusts such as delay and disruption tolerant networks (DTNs) and mobile opportunistic networking, can exploit social relationships between people in order to efficiently disseminate messages through such challenged networks. We believe that these types of networks better model and reflect human mobility pattern and limitations, and so are more naturally suited as a platform for tackling the problems mentioned above.

Management of Energy Consumption on Cluster Based Routing Protocol for MANET

The usage of light-weight mobile devices is increasing rapidly, leading to demand for more telecommunication services. Consequently, mobile ad hoc networks and their applications have become feasible with the proliferation of light-weight mobile devices. Many protocols have been developed to handle service discovery and routing in ad hoc networks. However, the majority of them did not consider one critical aspect of this type of network, which is the limited of available energy in each node. Cluster Based Routing Protocol (CBRP) is a robust/scalable routing protocol for Mobile Ad hoc Networks (MANETs) and superior to existing protocols such as Ad hoc On-demand Distance Vector (AODV) in terms of throughput and overhead. Therefore, based on this strength, methods to increase the efficiency of energy usage are incorporated into CBRP in this work. In order to increase the stability (in term of life-time) of the network and to decrease the energy consumption of inter-cluster gateway nodes, an Enhanced Gateway Cluster Based Routing Protocol (EGCBRP) is proposed. Three methods have been introduced by EGCBRP as enhancements to the CBRP: improving the election of cluster Heads (CHs) in CBRP which is based on the maximum available energy level, implementing load balancing for inter-cluster traffic using multiple gateways, and implementing sleep state for gateway nodes to further save the energy. Furthermore, we propose an Energy Efficient Cluster Based Routing Protocol (EECBRP) which extends the EGCBRP sleep state concept into all idle member nodes, excluding the active nodes in all clusters. The experiment results show that the EGCBRP decreases the overall energy consumption of the gateway nodes up to 10% and the EECBRP reduces the energy consumption of the member nodes up to 60%, both of which in turn contribute to stabilizing the network.