Low-power Mobile Devices Research Articles

On energy complexity of fully-connected layers

The massive increase in the size of deep neural networks (DNNs) is accompanied by a significant increase in energy consumption of their hardware implementations which is critical for their widespread deployment in low-power mobile devices. In our previous work, an abstract hardware-independent model of energy complexity for convolutional neural networks (CNNs) has been proposed and experimentally validated. Based on this model, we provide a theoretical analysis of energy complexity related to the computation of a fully-connected layer when its inputs, outputs, and weights are transferred between two kinds of memories (DRAM and Buffer). First, we establish a general lower bound on this energy complexity. Then, we present two dataflows and calculate their energy costs to achieve the corresponding upper bounds. In the case of a partitioned Buffer, we prove by the weak duality theorem from linear programming that the lower and upper bounds coincide up to an additive constant, and therefore establish the optimal energy complexity. Finally, the asymptotically optimal quadratic energy complexity of fully-connected layers is experimentally validated by estimating their energy consumption on the Simba and Eyeriss hardware.

Energy Complexity of Convolutional Neural Networks.

The energy efficiency of hardware implementations of convolutional neural networks (CNNs) is critical to their widespread deployment in low-power mobile devices. Recently, a number of methods have been proposed for providing energy-optimal mappings of CNNs onto diverse hardware accelerators. Their estimated energy consumption is related to specific implementation details and hardware parameters, which does not allow for machine-independent exploration of CNN energy measures. In this letter, we introduce a simplified theoretical energy complexity model for CNNs, based on only a two-level memory hierarchy that captures asymptotically all important sources of energy consumption for different CNN hardware implementations. In this model, we derive a simple energy lower bound and calculate the energy complexity of evaluating a CNN layer for two common data flows, providing corresponding upper bounds. According to statistical tests, the theoretical energy upper and lower bounds we present fit asymptotically very well with the real energy consumption of CNN implementations on the Simba and Eyeriss hardware platforms, estimated by the Timeloop/Accelergy program, which validates the proposed energy complexity model for CNNs.

Deployment of Offline Learning Management Systems: Comparing the Performance of Selected Micro-servers in Tanzania

Low-powered mobile devices such as Raspberry Pis and tablets can be used as micro-servers to implement offline Learning Management Systems (LMS). Despite their potential, especially for low-income countries, such as Tanzania, no research is available detailing the affordances of these devices for supporting LMS features. This study investigated the suitability of various low-cost micro-servers for deploying LMSs. It compared the performance of the Raspberry Pi, Android tablet, and Chromebook in terms of LMS benchmarking, response time, and resource utilisation. Results showed all devices had sufficient hardware resources to support the LMS, however, software stacks, I/O performance, and platform optimisations affected the micro-servers' performance. The Chromebook had the best performance in terms of response time, followed by the Raspberry Pi and tablets. In terms of cost, the Raspberry PI was the cheapest option. The installation process for tablets was more cumbersome than the other devices, meaning the devices with better tooling and a more conventional software stack were a better option for deploying offline micro-servers.

Tailored Hidden Markov Model: A Tailored Hidden Markov Model Optimized for Cellular-Based Map Matching

Although the GPS-based positioning is ubiquitous for its high precision, the high power consumption brought by the high sampling frequency and the poor GPS signal penetration limits its availability in locating low-power mobile devices (especially mobile phones). As a promising complement, the cellular-based positioning has attracted great attention since it consumes much less power as well as its higher availability. However, the sparsity of cellular-based data (due to lower sampling rate) and large localization errors make the measurement accuracy becomes the main challenge of the cellular-based positioning. hidden Markov model can well solve the problem of positioning error of GPS data, but it is less accurate when applied to map matching of cellular-base data. Therefore, to improve accuracy, in this article, we propose a novel algorithm called the tailored hidden Markov model (THMM) that is optimized for the cellular-based data. Specifically, the geometric, the topological, and the probabilistic characteristics have been considered and fully exploited in the THMM design. Our proposed schemes are evaluated using real-world motor vehicle movement trajectories collected in Tianjin and the experimental results are encouraging compared with the state-of-the-art algorithms.

Augmented Reality and GPS-Based Resource Efficient Navigation System for Outdoor Environments: Integrating Device Camera, Sensors, and Storage

Contemporary navigation systems rely upon localisation accuracy and humongous spatial data for navigational assistance. Such spatial-data sources may have access restrictions or quality issues and require massive storage space. Affordable high-performance mobile consumer hardware and smart software have resulted in the popularity of AR and VR technologies. These technologies can help to develop sustainable devices for navigation. This paper introduces a robust, memory-efficient, augmented-reality-based navigation system for outdoor environments using crowdsourced spatial data, a device camera, and mapping algorithms. The proposed system unifies the basic map information, points of interest, and individual GPS trajectories of moving entities to generate and render the mapping information. This system can perform map localisation, pathfinding, and visualisation using a low-power mobile device. A case study was undertaken to evaluate the proposed system. It was observed that the proposed system resulted in a 29 percent decrease in CPU load and a 35 percent drop in memory requirements. As spatial information was stored as comma-separated values, it required almost negligible storage space compared to traditional spatial databases. The proposed navigation system attained a maximum accuracy of 99 percent with a root mean square error value of 0.113 and a minimum accuracy of 96 percent with a corresponding root mean square value of 0.17.

A Real-Time Semantic Segmentation Approach for Autonomous Driving Scenes

<p indent=0mm>An important part of autonomous driving is the perception of the driving environment of the car, which has created a strong demand for high precision semantic segmentation algorithms that can be run in real time on low-power mobile devices. However, when analyzing the factors that affect the accuracy and speed of the semantic segmentation network, it can be found that in the structure of the previous semantic segmentation algorithm, spatial information and context features are difficult to take into account at the same time, and using two networks to obtain spatial information and context information separately will increase the amount of calculation and storage. Therefore, a new structure is proposed that divides the spatial path and context path from the network based on the residual structure, and a two-path real-time semantic segmentation network is designed based on this structure. The network contains a feature fusion module and an attention refinement module, which are used to realize the function of fusing the multi-scale features of two paths and optimizing the output results of context path. The network is based on the PyTorch framework and uses NVIDIA 1080Ti graphics cards for experiments. On the road scene data set Cityscapes, mIoU reached 78.8%, and the running speed reached 27.5 fps.

Catalytic device for environmentally friendly combustion of liquid fuels on the base of structured glass-fiber catalyst

The article presents the results of the development and testing of a prototype device for the combustion of liquid fuels and the disposal of liquid toxic waste based on structured microfiber catalysts. The technology is based on a two-stage approach combining homogeneous and catalytic oxidation, which is carried out in a structured spiral cartridge with a modified glass fiber platinum-containing catalyst IC-12-S111. A prototype of a catalytic device with a thermal power of 18 kW was created and tested on a model fuel (diesel fuel). The optimal combustion mode is realized at a fuel consumption of 1.8 L/h, an air flow supplied to the burner - 12.3 st.m3/h and to the mixing zone in front of the catalytic block - 67 st.m3/h. At the same time, the air leaving the device contains practically no toxic impurities (CO, nitrogen and sulfur oxides). An important advantage of the proposed technology is its modularity, that is, the possibility of implementation in the form of low-power mobile devices, from which it is possible to form transportable and quickly deployable systems of any total capacity and a wide range of purposes.

Memory access minimization for mean-shift tracking in mobile devices

Due to the development of artificial intelligence and computer vision technology, many autonomous drones have been studied. However, computer vision technology requires high performance CPU due to its high complexity, and battery consumption is so high that drones are constrained to fly for a long time. Therefore, low-power mobile devices require tracking algorithms that minimize battery consumption. In this paper, we propose a mean-shift based tracking algorithm that minimizes memory access to reduce battery consumption. To accomplish this, we minimize the number of memory accesses by using an algorithm that divides the direction of the mean-shift vector into eight, and calculates the sum of the density maps only for the new area without calculating the sum of the density maps for the already calculated area. It is possible to increase the calculation efficiency by lowering the memory access cost. Experimental results show that the proposed method is more efficient than the existing method.

Improved authentication method in embedded networks systems. An autonomous vehicle approach

By adding connectivity to in-vehicle networks (including multimedia devices) and external networks (e.g. wireless and Internet) the attack surface was dramatically extended. In this context, there are several types of attacks already been demonstrated on automotive/AV control networks using compromised network connection or physical manipulation. Subsequently, the success attacks violate safety requirements, being able to disrupt system operation or even to take over operational control. In order to avoid false authentication or identity theft of devices or IoT, this paper proposes a time-based authentication method. The proposed method increases the degree of cybersecurity and allows its implementation on independent, low power mobile devices. Finally, a critical conclusion regarding the proposed authentication methods is presented.

SMCP: a Secure Mobile Crowdsensing Protocol for fog-based applications

The possibility of performing complex data analysis through sets of cooperating personal smart devices has recently encouraged the definition of new distributed computing paradigms. The general idea behind these approaches is to move early analysis towards the edge of the network, while relying on other intermediate (fog) or remote (cloud) devices for computations of increasing complexity. Unfortunately, because both of their distributed nature and high degree of modularity, edge-fog-cloud computing systems are particularly prone to cyber security attacks that can be performed against every element of the infrastructure. In order to address this issue, in this paper we present SMCP, a Secure Mobile Crowdsensing Protocol for fog-based applications that exploit lightweight encryption techniques that are particularly suited for low-power mobile edge devices. In order to assess the performance of the proposed security mechanisms, we consider as case study a distributed human activity recognition scenario in which machine learning algorithms are performed by users’ personal smart devices at the edge and fog layers. The functionalities provided by SMCP have been directly compared with two state-of-the-art security protocols. Results show that our approach allows to achieve a higher degree of security while maintaining a low computational cost.

Ultralow Power Nonvolatile Logic Based on Spin-Orbit and Exchange Coupled Nanowires

The rapid growth of Internet of Things and emphasis on low power mobile devices has inspired researchers to explore newer devices, such as nonvolatile logic devices. Due to their miniaturized dimensions and zero leakage, spintronic devices are among the most promising devices in literature. In this paper, we propose a new logic style based on spin-orbit and exchange coupled nanowires. Our logic style takes advantage of two key features, namely, spin-orbit coupling and exchange coupling fields to enhance energy efficiency. Furthermore, we demonstrate dedicated input and output paths for the proposed device resulting in inherent asymmetry and simplified operation. Furthermore, thanks to our device structure, fringing dipole fields between devices are eliminated leading to improved area utilization and smaller area. In order to evaluate our proposed logic style, a simulation framework was developed. This simulation framework consists of a micromagnetic dynamics module, electron transport module, and a circuit integrator. Using our simulation framework, we demonstrate 88% reduction in power delay product and 93% reduction in the delay and 37% reduction in the area for an inverter compared to previously logic styles based on motion of domain walls.

Privacy Protection for Telecare Medicine Information Systems with Multiple Servers Using a Biometric-based Authenticated Key Agreement Scheme

Telecare medical information systems (TMIS) allow patients remotely login medical service providers to acquire their medical information and track their health status through unsecured public networks. Hence, the privacy of patients is vulnerable to various types of security threats and attacks, such as the leakage of medical records or login footprints and the forgery attacks. Many anonymous three-factor authentication and key agreement (AKA) schemes have been proposed for TMIS with single server, but none of them is suited for TMIS with multiple servers. In this paper, we propose a biometric-based three-factor AKA scheme to protect user anonymity and untraceability in TMIS with multiple servers. We will construct a security model of a three-factor AKA scheme with user anonymity in TMIS with multiple servers, and give a formal security proof of the proposed scheme. The security of the proposed scheme is based on the elliptic curve decisional Diffie-Hellman problem assumption and hash function assumption. We will show that the proposed scheme is efficient enough for low-power mobile devices.

A heterogeneous user authentication and key establishment for mobile client–server environment

In a mobile client–server environment, a low-power mobile device wants to access a strong server to get some kind of services. User authentication and key establishment are two basic security requirements for this environment. Without the user authentication, an unauthorized user can access the server and gets the services. Without the key establishment, the communication between the user and the server will be disclosed. Recently, some user authentication and key establishment protocols were designed. However, all of them are homogeneous since the client and the server belong to the same cryptosystem. That is, both the client and the server belong to public key infrastructure or identity-based cryptosystem or self-certified cryptosystem. Such design does not comply with the characteristic of mobile client–server application. In this paper, we design a heterogeneous user authentication and key establishment protocol using a signcryption scheme. In this protocol, the client uses identity-based cryptosystem and the server uses the public key infrastructure. As compared with existing works, our protocol has the lowest cost in computation and communication.

A Novel Negative Location Collection Method for Finding Aggregated Locations

Currently, many intelligent transportation systems (ITSs) require aggregation of location information. Although users enjoy the convenience of ITSs, privacy concerns have resulted in user’s caution in offering location information. Certain studies have been performed to preserve user location privacy. However, most studies that focus on preserving location privacy require a trusted third party or do not consider the movements of users. In this paper, we propose a method based on negative surveys that can be used to estimate the number of people in geographic locations. This method, which can preserve user privacy regardless of user movements, adopts a simple negative survey algorithm for user devices and an estimation algorithm for the server and might be suitable for low-power mobile devices. The experimental results demonstrate that our method is capable of locating the people’s gathering places with fine control granularity, which renders it a promising application.

Secure transmission of multimedia contents over low-power mobile devices

Secure transmission of multimedia contents is computationally challenging due to the sheer volume of data involved, and achieving this by fully encrypting multimedia contents is not a viable solution for mobile devices, as they run on limited battery power and employ relatively slower processors compared to their desktop counterparts. Also, in some particular application scenarios, the value, secrecy, and/or privacy of the information is time dependent and short lived. This paper presents a highly scalable approach for securing multimedia streams on mobile devices by encrypting intelligently only selected portions of the bit stream. More importantly, the proposed solution works mainly on compressed bit stream and does not require any media decoding. It encrypts on the average less than 3% of a packet load and provides robust security equivalent to that of a fully encrypted bit stream. This solution has been implemented on desktop, laptop, netbook and Nokia N series platforms. The proposed scheme is approximately 15 times faster (on average) and processes 8 to 12 times more information for a given battery life when compared to a state of the art full encryption based solutions.

An Effective Authentication Scheme Using DCT for Mobile Devices

This paper proposes an image authentication scheme for mobile devices. The proposed scheme generates an image watermark by using discrete cosine transform (DCT) and hides the watermark in the spatial pixels for image authentication and tamper detection. The hiding operator used in this paper is very simple in a mobile environment allowing high-speed authentication using a low-power mobile device. The quality of the stego-image and the recovered image becomes excellent as a result of the proposed scheme.

Investigation of Next Generation Internet Protocol Mobility-Assisted Solutions for Low Power and Lossy Networks

With the phenomenal increase of computing devices, especially the low powered mobile devices created many new requirements for traditional networking protocols and standards. The IPv6 was introduced as the solution for addressing requirements but still, some major issues persist due to the limitations of computing devices such as power related limitations, the challenge in the deployment environments and restrictions allied to computing and memory capacity. Many new standards were proposed such as the IPv6 over Low-Power Wireless Personal Area Networks (6LoWPAN), the IPv6 routing protocol for Low Power and Lossy Networks (RPL) to meet the challenges. Even though RPL greatly satisfies the requirements for low-powered networks, quite a few issues remain open for enhancement, especially with respect to mobility support. This present paper examines next-generation internet protocol mobility schemes for the RPL and 6LoWPAN and analyzes how the proposed variants satisfy the necessities for low power and lossy networks.

Optimization of intrinsic layer thickness, dopant layer thickness and concentration for a-SiC/a-SiGe multilayer solar cell efficiency performance using Silvaco software

Solar cell is expanding as green renewable alternative to conventional fossil fuel electricity generation, but compared to other land-used electrical generators, it is a comparative beginner. Many applications covered by solar cells starting from low power mobile devices, terrestrial, satellites and many more. To date, the highest efficiency solar cell is given by GaAs based multilayer solar cell. However, this material is very expensive in fabrication and material costs compared to silicon which is cheaper due to the abundance of supply. Thus, this research is devoted to develop multilayer solar cell by combining two different layers of P-I-N structures with silicon carbide and silicon germanium. This research focused on optimising the intrinsic layer thickness, p-doped layer thickness and concentration, n-doped layer thickness and concentration in achieving the highest efficiency. As a result, both single layer a-SiC and a-SiGe showed positive efficiency improvement with the record of 27.19% and 9.07% respectively via parametric optimization. The optimized parameters is then applied on both SiC and SiGe P-I-N layers and resulted the convincing efficiency of 33.80%.

SURE: A Novel Approach for Self Healing Battery Starved Users Using Energy Harvesting

Radio Frequency (RF) energy harvesting holds a promising future for energizing low power mobile devices in next generation wireless networks. Harvesting from a dedicated RF energy source acquires much more energy than simply harvesting from ambient RF sources. In this paper, novel Self-healing of Users equipment by RF Energy transfer scheme is introduced between the network operator and battery starved users to heal and extend their battery life time by sending dedicated energy from different sources in order to be aggregated and harvested by starved users. This approach depends on the concept of Energy as a Service where the network operator delivers energy to battery starved users in the next generation networks. A mixed integer non-linear optimization problem is formulated and solved efficiently using three heuristic algorithms. Simulation results prove that sufficient amounts of energy can be delivered to starved users while minimizing their uplink power requirements and guaranteeing a minimum uplink data rate.

The achievement of the super short channel control in the magnetic Ge n-FinFETs with the negative capacitance effect

Super short channel control (sub-threshold swing = 95 mV/dec) in Ge n-type Fin Field-Effect Transistors (n-FinFETs) is achieved through the promising gate stack characteristics of gate leakage (Jg)-equivalent-oxide-thickness (EOT) and ultra-high k-value (∼312) with the implement of the magnetic gate stack scheme. On the other hand, the negative capacitance effect is also observed in the magnetic Ge n-FinFETs by the extraction of the body factor (m = 0.47) at low temperature measurement. The proposed magnetic gate stack scheme (tetragonal-phase BaTiO3 as the dielectric layer and magnetic FePt film as the metal gate) in the Ge n-FinFETs has super Jg-EOT characteristics, negative capacitance phenomenon, and promising transistor performance. It provides the useful solution for the future low power mobile devices designed on the high mobility (Ge) material.