Ubiquitous Sensor Research Articles

Use of Mobile Crowdsensing in Disaster Management: A Systematic Review, Challenges, and Open Issues.

With the increasing efforts to utilize information and communication technologies (ICT) in disaster management, the massive amount of heterogeneous data that is generated through ubiquitous sensors paves the way for fast and informed decisions in the case of disasters. Utilization of the big "sensed" data leads to an effective and efficient management of disaster situations so as to prevent human and economic losses. The advancement of built-in sensing technologies in smart mobile devices enables crowdsourcing of sensed data, which is known as mobile crowdsensing (MCS). This systematic literature review investigates the use of mobile crowdsensing in disaster management on the basis of the built-in sensor types in smart mobile devices, disaster management categories, and the disaster management cycle phases (i.e., mitigation, preparedness, response, and recovery activities). Additionally, this work seeks to unveil the frameworks or models that can potentially guide disaster management authorities towards integrating crowd-sensed data with their existing decision-support systems. The vast majority of the existing studies are conceptual as they highlight a challenge in experimental testing of the disaster management solutions in real-life settings, and there is little emphasis on the use cases of crowdsensing through smartphone sensors in disaster incidents. In light of a thorough review, we provide and discuss future directions and open issues for mobile crowdsensing-aided disaster management.

Trust assessment for mobile crowdsensing via device fingerprinting

Mobile crowdsensing leverages the ubiquitous sensors of smart devices to facilitate various sensing applications. Users who participate in contributing data usually get rewards from the task requester, while there is a potential risk that someone would preempt the task and provide a forged sensing report for seeking revenue with minimal effort. Thus, trust assessment is essential to identify those irregular sensing reports. The existing methods mainly consider users’ reputations and estimate the trustworthiness upon the difference from the aggregated result. However, they still face a severe problem when a majority of reports are invalid or low-quality caused by the repeated submission, e.g., a user can switch multiple accounts on a single device to repeatedly submit forged reports. To tackle this problem, we design a trust assessment scheme with an enhanced device fingerprinting algorithm. Briefly, to reduce the influence of the repeated sensing reports, we first compute their unique fingerprints derived from the intrinsic characteristics of sensors and assign an initial trust weight for each report. Then, to improve the accuracy of the assessment, we further compute the similarity of the reports to obtain their final trust values. Extensive evaluations are conducted to justify the effectiveness of our proposed design.

Low temperature, area-selective atomic layer deposition of NiO and Ni

Nickel and nickel oxide are utilized within various device heterostructures for chemical sensing, solar cells, batteries, etc. Recently, the rising interest in realizing low-cost, flexible electronics to enable ubiquitous sensors and solar panels, next-generation displays, and improved human-machine interfaces has driven interest in the development of low-temperature fabrication processes for the integration of inorganic devices with polymeric substrates. Here, we report the low-temperature area-selective atomic layer deposition of Ni by reduction of preformed NiO. Area-selective deposition of NiO is performed at 100 °C using bis(N,N'-di-tert-butylacetamidinato) nickel(II) and water on SiO2 and polystyrene. NiO grows two-dimensionally and without nucleation delay on oxide substrates but not on SiNx or polystyrene, which require surface treatments to promote NiO nucleation. Additionally, prepatterned sp2 carbon-rich resists inhibit the nucleation of NiO, and in this way, carbon-free NiO may be patterned. Subsequent thermal reduction of NiO to Ni was investigated using H2 (50–80 m Torr) and thermally generated H-atoms (3 × 10−5 Torr chamber pressure). Due to the relatively high free surface energy of Ni metal, Ni films undergo dewetting at elevated temperatures when solid-state transport is enabled. Reduction of NiO to Ni is demonstrated at 100 °C and below using atomic hydrogen. In situ x-ray photoelectron spectroscopy is used to determine oxidation state and ex situ x-ray reflectivity and atomic force microscopy are used to probe the film thickness and surface morphology, respectively.

Protic Ionic Liquid Electrolytes to Increase Areal Energy Density of RuO2 Micro-Supercapacitors

The rising growth of smart and autonomous microelectronic devices in the IoT (Internet of Things) era urges the development of advanced microscale energy sources with tailor-made features and customized energy/power requirements [1]. Micro-supercapacitors (MSCs) emerged as potential energy storage devices complementing micro-batteries to power ubiquitous sensor networks needed to foster the development of IoT. However, the low cell voltage and low energy density remain major bottleneck that prevents their application at a large scale in real devices. To mitigate this issue, several studies have been devoted to the engineering of MSC electrode materials and structural architecting of current collectors to enhance the surface area and areal energy density by considering the limited available footprint area [2]. This approach however has associated challenges such as complex synthesis route, the deleterious interfacial and mechanical stability of the electrode, and compatibility issues with the electrolyte and the current collector underneath [3]. Another important challenge to solve for reaching high energy density values in MSCs is the limited electrochemical stability window (ESW) of the electrolytes used as energy stored is directly related to the square of the cell voltage [4]. The electrolytes play a major role in deciding the ESW and liquid-state electrolytes currently employed are troublesome for the microfabrication process due to leakage, evaporation, and safety issues. Therefore, it’s imperative to develop alternative electrolytes including solid-state electrolytes reconcilable to the target application of MSCs. To address the low energy density challenge of current MSCs, we have developed interdigitated MSCs using hydrous ruthenium dioxide (RuO2) electrodes in combination with novel protic ionic liquid (PIL)-based electrolytes able to provide pseudocapacitance while affording a higher ESWs as compared to conventional aqueous electrolytes. As a state-of-the-art pseudocapacitive electrode material, RuO2 owns the key merits of excellent conductivity, high electrochemical reversibility, and cycling stability [5], whereas PILs could help alleviate issues facing currently used electrolytes such as evaporation and encapsulation problems pertaining to aqueous-based and flammability of common organic electrolytes [6], [7], [8]. In the next step, the slow proton transport kinetics of PILs were addressed by the doping of silicotungstic acid (SiWa, H4SiW12O40) with the PIL, which further boosted the pseudocapacitive current response with enlarged cell voltage. The real MSC device was realized by the use of RuO2 deposited on interdigitated porous Au current collectors having a high area enlargement factor (AEF) in combination with triethylammonium bis(trifluoromethanesulfonyl)imide (TEAH-TFSI)-based PIL. The resultant 3D MSC rendered a cell voltage exceeding 2V with areal capacitance as high as 86 mF cm-2 at 5 mV s-1 on par with the performance of 3D MSC tested using 0.5 M H2SO4 (cell voltage of 0.9 V and areal capacitance of 85 mF cm-2 at 5 mV s-1) but higher energy density performance (more than 4 times) using similar number of RuO2 deposition cycles. To demonstrate the potential integration in real on-chip device application, ionogel-based all-solid-state MSC is developed that showed performance comparable to liquid-state electrolyte with superior long-term cycling stability. This study gives a new perspective to develop all-solid-state micro-supercapacitors using pseudocapacitive active materials that can operate in ionic-liquid-based non-aqueous electrolytes compatible with on-chip IoT-based device applications seeking high areal energy/ power performance.

Investigation of protic ionic liquid electrolytes for porous RuO2 micro-supercapacitors

The rapid advancement of the Internet of things (IoT) with applications across various sectors urges the development of miniaturized energy-storage devices that can harvest or deliver energy with high power capabilities. While micro-supercapacitors can meet the high-power requirements of ubiquitous sensors connected to IoT networks, their low voltage and low energy density remain a major bottleneck preventing their wide-scale adoption. In this report, we develop micro-supercapacitors using RuO2 electrodes providing pseudocapacitive charge storage in protic ionic liquid-based non-aqueous electrolytes while enlarging their operational voltage. The triethylammonium bis(trifluoromethanesulfonyl)imide (TEAH-TFSI)-based interdigitated porous RuO2 micro-supercapacitors showed an extended cell voltage up to 2 V with 4 times more energy density compared with conventional H2SO4 electrolyte. We then developed an all-solid-state micro-supercapacitor using TEAH-TFSI-based ionogel electrolyte able to deliver high areal capacitance (79 mF cm−2 at 2 mV s−1) and long-term cycling stability that is superior to state-of-the-art ionogel-based micro-supercapacitors employing carbon-based or pseudocapacitive materials. This study gives a new perspective to develop all-solid-state micro-supercapacitors using pseudocapacitive active materials that can operate in ionic-liquid-based non-aqueous electrolytes compatible with on-chip IoT-based device applications seeking high areal energy/power performance.

Machine learning model for snow depth estimation using a multisensory ubiquitous platform

Snow depth estimation is an important parameter that guides several hydrological applications and climate change prediction. Despite advances in remote sensing technology and enhanced satellite observations, the estimation of snow depth at local scale still requires improved accuracy and flexibility. The advances in ubiquitous and wearable technology promote new prospects in tackling this challenge. In this paper, a wearable IoT platform that exploits pressure and acoustic sensor readings to estimate and classify snow depth classes using some machine-learning models have been put forward. Significantly, the results of Random Forest classifier showed an accuracy of 94%, indicating a promising alternative in snow depth measurement compared to in situ, LiDAR, or expensive large-scale wireless sensor network, which may foster the development of further affordable ecological monitoring systems based on cheap ubiquitous sensors.

All-in-one, bio-inspired, and low-power crypto engines for near-sensor security based on two-dimensional memtransistors

In the emerging era of the internet of things (IoT), ubiquitous sensors continuously collect, consume, store, and communicate a huge volume of information which is becoming increasingly vulnerable to theft and misuse. Modern software cryptosystems require extensive computational infrastructure for implementing ciphering algorithms, making them difficult to be adopted by IoT edge sensors that operate with limited hardware resources and at low energy budgets. Here we propose and experimentally demonstrate an “all-in-one” 8 × 8 array of robust, low-power, and bio-inspired crypto engines monolithically integrated with IoT edge sensors based on two-dimensional (2D) memtransistors. Each engine comprises five 2D memtransistors to accomplish sensing and encoding functionalities. The ciphered information is shown to be secure from an eavesdropper with finite resources and access to deep neural networks. Our hardware platform consists of a total of 320 fully integrated monolayer MoS2-based memtransistors and consumes energy in the range of hundreds of picojoules and offers near-sensor security.

Video Analytics in Elite Soccer: A Distributed Computing Perspective.

Ubiquitous sensors and Internet of Things (IoT) technologies have revolutionized the sports industry, providing new methodologies for planning, effective coordination of training, and match analysis post game. New methods, including machine learning, image and video processing, have been developed for performance evaluation, allowing the analyst to track the performance of a player in real-time. Following FIFA's 2015 approval of electronics performance and tracking system during games, performance data of a single player or the entire team is allowed to be collected using GPS-based wearables. Data from practice sessions outside the sporting arena is being collected in greater numbers than ever before. Realizing the significance of data in professional soccer, this paper presents video analytics, examines recent state-of-the-art literature in elite soccer, and summarizes existing real-time video analytics algorithms. We also discuss real-time crowdsourcing of the obtained data, tactical and technical performance, distributed computing and its importance in video analytics and propose a future research perspective.

Wi-Fi Can Do More: Toward Ubiquitous Wireless Sensing

The next big deal for Wi-Fi is not about communication and networking, but sensing. Wireless sensing technology is turning a Wi-Fi device into a ubiquitous sensor, which not only adds a brand new dimension to the functions, capabilities, and applications of all Wi-Fi systems, but also revolutionizes how sensing, especially human-centric sensing, is practiced. Wi-Fi sensing utilizes ambient Wi-Fi signals to analyze and interpret human and object movements, underpinning many sensing applications including motion sensing, sleep monitoring, and fall detection, among others. These new sensing functionalities can benefit the global Wi-Fi ecosystem including integrated circuit manufacturers, device manufacturers, system integrators, application developers, and ultimately end users. In this article, we introduce the concepts, principles, and challenges of Wi-Fi sensing, and share our unique technologies that have been massively deployed for real-world applications. We foresee that Wi-Fi sensing will enter billions of devices and millions of homes, creating a smarter space for a smarter life. Today is just the beginning of this revolution.

67‐2: Invited Paper: Organic Thin‐Film Transistor Flexible Hybrid Integration for Low‐power Ubiquitous Sensor Systems

This paper firstly introduces a flexible hybrid integration (FHI) architecture based on organic thin‐film transistors (OTFTs) for ubiquitous sensor systems, followed by description of the low power OTFT technologies. Then, progresses on sensor integration based on the OTFTs for FHI are discussed, proving the versatility of the technology.

Characterizing User Experiences With an SMS Text Messaging-Based mHealth Intervention: Mixed Methods Study.

BackgroundLimited access to mental health care services due to provider shortages, geographic limitations, and cost has driven the area of mobile health (mHealth) care to address these access gaps. Reports from the Cohen Veterans Network and National Council for Behavioral Health show that in states where mental health care is more accessible, 38% of people still do not receive the care they need. mHealth strategies help to provide care to individuals experiencing these barriers at lower cost and greater convenience, making mHealth a great resource to bridge the gaps.ObjectiveWe present a mixed methods study to evaluate user experiences with the mental mHealth service, Cope Notes. Specifically, we aimed to investigate the following research questions: How do users perceive the service in relation to stigma, impact of the intervention, and perceived usefulness? How do users rate the Cope Notes service and SMS text messaging along various dimensions of acceptability? What is the relationship between Cope Notes SMS text message ratings, user personality, and coping strategies? What are user perspectives of leveraging ubiquitous sensing technologies to improve delivery and provide tailored content?MethodsWe performed qualitative interviews with Cope Notes users (N=14) who have used the service for at least 30 days to evaluate their experiences and usefulness of the service. These interviews were coded by 2 raters (SLK and JL), and the interrater reliability was calculated with SPSS (IBM Corp) at 61.8%. In addition, participants completed quantitative measures, including a user experiences survey, personality inventory (Big Five Inventory-10), and coping assessment (Brief Coping Orientation to Problems Experienced).ResultsWe derived 7 themes from our qualitative interviews: Likes or Perceived Benefits, Dislikes or Limitations, Suggested Changes, Stigma or Help Seeking, Perceptions of Ubiquitous Sensing, Cultural Sensitivity, and Alternative mHealth Resources. Exploratory analyses between acceptability ratings of Cope Notes and personality factors showed statistically significant positive relationships between seeing oneself as someone who is generally trusting and acceptability items, the most significant being item 7 (I fully understood the sentiment behind Cope Notes Messages) with (rs(10)=0.82, P=.001). We also found statistically significant relationships between acceptability and Brief Coping Orientation to Problems Experienced items, with the strongest positive correlation between participants strongly endorsing coping by accepting the reality that an event has happened and acceptability item 7 (rs(8)=0.86, P=.001).ConclusionsOur study found that Cope Notes subscribers appreciate the service for reframing their mental wellness with statistically significant correlations between personality and acceptability of the service. We found that some users prefer a more personalized experience with neutral to positive reactions to a potential companion app that continuously monitors user behavior via smartphone sensors to provide just-in-time interventions when users need it most.

Internet of Things (IoT) Based Activity Recognition Strategies in Smart Homes: A Review

A smart home, which is an extension of a traditional home, is equipped with ubiquitous sensors embedded in consumer appliances, connected via sensing technologies such as radio frequency, and communicating through the internet to gather information. They often communicate using appropriate protocols such as MQTT, CoAP, or HTTP to ensure the smooth transmission of data used by a variety of smart home services. Human activity recognition is one of the services provided by this IoT method of data collection from the sensor network when activated by residents. The obtained data can be subjected to extensive preprocessing and feature extraction tasks before being learned using appropriate machine learning or deep learning algorithms to generate a model capable of managing human activities more effectively. This technique is challenged by the nature of IoT technology and perceived data, as well as by human differences, which necessitated additional processing tasks to select significant features for the learning algorithms. In this work, we focus our review on activity recognition implementation strategies by examining various sensors and sensing technologies used to collect useful data from IoT devices, reviewing preprocessing and feature extraction techniques, as well as classification algorithms used to recognize human activities in smart homes. Many relevant works were examined and their achievements compared.The research demonstrates that IoT sensor technology for recognizing human activity in a smart home is practically feasible and efficient, even with individual differences in the smart home. However, it was discovered to be susceptible to issues such as insufficient and imbalanced data, annotation scarcity, and computational complexity. Finally, the study suggests that associating sensor data from the Internet of Things with numerous labels of activities based on time can help decrease computing overhead and improve activity recognition.

Communication Technologies in Emergency Situations

Emergency situations such as wildfires, water floods, or even terrorist attacks require continuous communication between the coordination centres, the several on-the-field teams, and their respective devices to properly address the adverse circumstances. From a technological point of view, this can be best seen as a live Ubiquitous Sensor Network—composed of human beings (e.g., first responders, victims) and devices (e.g., drones, environmental sensors, radios)—with stringent and special communication requirements in terms of flexibility, mobility, reliability, bandwidth, heterogeneity, and speed of deployment. However, for this specific use case, most of the already deployed and well-known communication technologies (e.g., satellite, 4G/5G) might become unusable and hard to repair due to the associated effects of the disaster itself. The purpose of this paper is (1) to review the emergency communications challenges, (2) to analyse the existing surveys on technologies for emergency situations, (3) to conduct a more updated, extensive, and systematic review of the emergency communications’ technologies, and (4) to propose a heterogeneous communication architecture able to communicate between moving agents in harsh conditions. The proposed approach is conceived to link the relocating agents that constitute a Ubiquitous Sensor Network spanning a large-scale area (i.e., hundreds of square kilometres) by combining Near Vertical Incidence Skywave technologies with Drone-Based Wireless Mesh Networks. The conclusions derived from this research aim to set up the fundamentals of a rapidly deployable Emergency Communications System inspired by the Ubiquitous Sensor Network paradigm.

Internet of Things for Smart Cities

This study uses a methodology based on an in-depth literature assessment of international organisations to define the meaning of the term "smart" in relation to cities. It also acknowledges the key dimensions and characteristics that define a smart city. The various measures of urban smartness are examined in order to establish the need for a shared understanding of what makes a smart city, what its highlights are, and how it performs in comparison to traditional urban regions. Smart cities advocate for future environments in which ubiquitous sensors, data supply and sharing, and information mash-up allow for improved support of all aspects of (social) life in human settlements. As this vision develops, evolves, and is shaped by a variety of application situations and adoption perspectives, a common requirement for scalable, pervasive, and adaptable solutions emerges.

DoSRA: A Decentralized Approach to Online Edge Task Scheduling and Resource Allocation

With the proliferation of novel Internet of Things (IoT) mobile applications and advanced communication technologies, nowadays we are surrounded by ubiquitous sensors and smart devices. These smart IoT devices generate a large volume of data day and night at the edge of the network, create a huge demand for edge computing resources, and thus, promote the emergence of the multiaccess edge computing (MEC) paradigm. In MEC environments, IoT devices or mobile users are allowed to offload their computational tasks to nearby edge servers to overcome the limitation of local computing resources. Though edge servers could provide low-latency service with high-responsible computing capabilities, they are still facing many challenges posed by the limited hardware resources and diverse offloading requests. However, traditional approaches are usually based on the centralized architecture and batch-processing scheduling mode, which might lead to low efficiency and high communication overhead. Besides, they also lack the consideration of task diversity and priorities, which are crucial in real-world application scenarios. Thus, smart task scheduling and resource provision strategies with a high real-time property are urgently needed for better user experience and higher resource utilization. In this article, we target the online edge IoT task scheduling and resource allocation problem and propose a decentralized approach (DoSRA). The experiments based on real-world edge environments have demonstrated that the proposed approach could achieve at most a 35.34% reduction on the average weighted offloading response time.

Radio-transparent dipole antenna based on a metasurface cloak

Antenna technology is at the basis of ubiquitous wireless communication systems and sensors. Radiation is typically sustained by conduction currents flowing around resonant metallic objects that are optimized to enhance efficiency and bandwidth. However, resonant conductors are prone to large scattering of impinging waves, leading to challenges in crowded antenna environments due to blockage and distortion. Metasurface cloaks have been explored in the quest of addressing this challenge by reducing antenna scattering. However, metasurface-based designs have so far shown limited performance in terms of bandwidth, footprint and overall scattering reduction. Here we introduce a different route towards radio-transparent antennas, in which the cloak itself acts as the radiating element, drastically reducing the overall footprint while enhancing scattering suppression and bandwidth, without sacrificing other relevant radiation metrics compared to conventional antennas. This technique opens opportunities for cloaking technology, with promising features for crowded wireless communication platforms and noninvasive sensing.

Rate Control Optimization Algorithm of Ubiquitous Power Sensor Network Tester Based on Multimedia

With more and more video application scenarios and increasing complexity, the popularization of multimedia information and the urgent need to update wireless sensor networks have led to the emergence and development of wireless multimedia sensor networks. This paper focuses on the rate control optimization algorithm of multimedia-based ubiquitous power sensor network, which is mainly studied from three aspects: constructing algorithm verification platform based on multimedia technology, performance analysis, and optimization of rate control optimization algorithm based on multimedia technology. An algorithm verification platform is established based on multimedia technology. This article uses a scalable video coding method to design a client-driven scalable video transmission system. The simulation results show that the algorithm can effectively improve the video quality. The algorithm of rate control based on multimedia technology is proposed. A reliable multipath routing algorithm based on AODV is proposed. Based on the energy balance algorithm, the node-independent routing mechanism and link quality estimation model are introduced. An independent and reliable active/standby communication link to the node of the destination node is established. Linux, the full name of GNU/Linux, is a free-to-use and freely disseminated Unix-like operating system. It is a POSIX-based multiuser, multitasking, multithreaded, and multi-CPU operating system. With the development of the Internet, Linux has gained support from software enthusiasts, organizations, and companies all over the world. In addition to maintaining a strong momentum of development in servers, it has made considerable progress in personal computers and embedded systems. Under the Linux platform, a large number of tests are performed and the performance of the control code rate is analyzed. The so-called sensor network is a distributed intelligent network system that is composed of a large number of small sensor nodes with wireless communication and computing capabilities deployed in the active area through a self-organizing method that can autonomously complete designated tasks according to the environment. The distance between the nodes of the sensor network is very short, and the multihop wireless communication method is generally used for communication. The sensor network can run in an independent environment, or it can be connected to the Internet through a gateway so that users can access it remotely. It is concluded that ABR is the most suitable rate control algorithm for video transmission. According to the CQP experiment results, the QP value set in the improved ABR algorithm is at most 40; the security from the construction of the ubiquitous power sensor network to the final monitoring data collection and upload process is verified by simulating the hypothesis attacks and faults. It is proved that the construction of ubiquitous power sensor network can reduce the energy consumption and delay of monitoring data collection and verify the superiority of constructing ubiquitous power sensor network.

Intelligent Sound Monitoring and Identification System Combining Triboelectric Nanogenerator‐Based Self‐Powered Sensor with Deep Learning Technique

AbstractUrban sound management is required in a variety of fields such as transportation, security, water conservancy and construction, among others. Given the diverse array of available noise sensors and the widespread opportunity to connect these sensors via mobile broadband Internet access, many researchers are eager to apply sound‐sensor networks for urban sound management. Existing sensing networks typically consist of expensive information‐sensing devices, the cost and maintenance of which limit their large‐scale, ubiquitous deployment, thus narrowing their functional measurement range. Herein, an innovative, low‐cost, sound‐driven triboelectric nanogenerator (SDTENG)‐based self‐powered sensor is proposed, from which the SDTENG is primarily comprised of fluorinated ethylene propylene membranes, conductive fabrics, acrylic shells, and Kapton spacers. The SDTENG‐based sensor has been integrated with a deep learning technique in the present study to construct an intelligent sound monitoring and identification system, which is capable of recognizing a suite of common road and traffic sounds with high classification accuracies of 99% in most cases. The novel SDTENG‐based self‐powered sensor combined with deep learning technique demonstrates a tremendous application potential in urban sound management, which will show the excellent application prospects in the field of ubiquitous sensor networks.

Deep unsupervised methods towards behavior analysis in ubiquitous sensor data

Behavioral analysis (BA) on ubiquitous sensor data is the task of finding the latent distribution of features for modeling user-specific characteristics. These characteristics, in turn, can be used for a number of tasks including resource management, power efficiency, and smart home applications. In recent years, the employment of topic models for BA has been found to successfully extract the dynamics of the sensed data. Topic modeling is popularly performed on text data for mining inherent topics. The task of finding the latent topics in textual data is done in an unsupervised manner. In this work we propose a novel clustering technique for BA which can find hidden routines in ubiquitous data and also captures the pattern in the routines. Our approach efficiently works on high dimensional data for BA without performing any computationally expensive reduction operations. We evaluate three different techniques namely Latent Dirichlet Allocation (LDA), the Non-negative Matrix Factorization (NMF), and the Probabilistic Latent Semantic Analysis (PLSA) for comparative study. We have analyzed the efficiency of the methods by using performance indices like perplexity and silhouette on three real-world ubiquitous sensor datasets namely, the Intel Lab, Kyoto, and MERL. Through rigorous experiments, we achieve silhouette scores of 0.7049 over the Intel Lab dataset, 0.6547 over the Kyoto dataset, and 0.8312 over the MERL dataset for clustering. In these cases, however, it is di cult to validate the results obtained as the datasets do not contain any ground truth information. Towards that, we investigate a self-supervised method that will be capable of capturing the inherent ground truths that are available in the dataset. We design a self-supervised technique which we apply on datasets containing ground truth and also without. We see that our performance on data without ground truth differs from that with ground truth by approximately 8% (F-score) hence showing the efficacy of self-supervised techniques towards capturing ground truth information.

Joint Optimization With DNN Partitioning and Resource Allocation in Mobile Edge Computing

With the rapid development of computing power and artificial intelligence, IoT devices equipped with ubiquitous sensors are gradually installed with intelligence. People can enjoy many conveniences with intelligent devices, such as face recognition, video understanding, and motion estimation. Currently, deep neural networks are the mainstream technology in intelligent mobile applications. Inspired by DNN model partition schemes, the paradigm of edge computing could be utilized collaboratively to improve the effectiveness of intelligent task execution in IoT devices. However, due to the dynamics of the wireless network environment and the increasing number of IoT devices, a DNN partition policy without adequate consideration would pose a significant challenge to the efficiency of task inference. Moreover, the shortage and high rental cost of edge computing resources make the optimization of DNN-based task execution more difficult. To cope with those situations, we propose a joint method by a self-adaptive DNN partition with cost-effective resource allocation to facilitate collaborative computation between IoT devices and edge servers. Our proposed online algorithm can be proved to ensure the overall rental cost within an upper bound above the optimal solution while guaranteeing the latency for DNN-based task inference. To evaluate the performance of our strategy, we conduct extensive trace-driven illustrative studies and show that the proposed method can achieve sub-optimal results and outperforms other alternative methods.