Internet of Things – ICIOT 2021

With the rapid development of the internet of things (IoT) and artificial intelligence (AI) technologies, human activity recognition (HAR) has been applied in a variety of domains such as security and surveillance, human-robot interaction, and entertainment. Even though a number of surveys and review papers have been published, there is a lack of HAR overview papers focusing on healthcare applications that use wearable sensors. Therefore, we fill in the gap by presenting this overview paper. In particular, we present our projects to illustrate the system design of HAR applications for healthcare. Our projects include early mobility identification of human activities for intensive care unit (ICU) patients and gait analysis of Duchenne muscular dystrophy (DMD) patients. We cover essential components of designing HAR systems including sensor factors (e.g., type, number, and placement location), AI model selection (e.g., classical machine learning models versus deep learning models), and feature engineering. In addition, we highlight the challenges of such healthcare-oriented HAR systems and propose several research opportunities for both the medical and the computer science community.

Although deep learning algorithms can achieve high performance, deep models may not learn the right concepts and can easily overfit their training datasets. In the context of IoT devices, the problem is further exacerbated by three factors. First, traffic may be encrypted, allowing very little visibility into the activity of the endpoints. Second, devices with different models and manufacturers may exhibit very different behaviors. Finally, contrary to domains like computer vision or natural language processing, there is no well-accepted representation for the network data that characterizes IoT devices. In this work, we capture real network traffic from different environments, and we demonstrate that training models to detect specific classes of IoT devices (e.g., cameras) using state-of-the-art techniques can lead to overfitting, and very poor performance on independent datasets. However, we then show that by applying domain knowledge, one can manually define engineered features and train simple models (e.g., a decision tree) that achieve an F-1 score of 0.956 on an independent dataset. These results show the feasibility of training generalizable models, but at the same time, raise questions on how best to transform and represent the raw network data to train classifiers for other classes of IoT devices (e.g., hubs, motion sensors) while minimizing manual feature engineering. We elaborate on the challenges, drawing analogies with other fields such as natural language processing.

The Internet of Things provides users with a variety of services that enable and intelligent and automated living environment design. University campuses represent an invaluable opportunity to optimize this approach, and smart campus services are already functional in many universities. This article carried out a systematic mapping of the Internet of Things (IoT) scenario within universities. This mapping was guided by research questions which aimed at assessing application areas, technologies, architectures, benefits and drawbacks described throughout the literature. This study selected 451 articles from January 2015 to May 2020 of which 39 articles were used in this systematic mapping. This article noted that resource management has been the main application area of smart campus IoT developments, primarily with a focus on energy issues. Architectures and models to improve data collection, processing, and communication are predominant in the papers we have found. In addition to contributing to researchers in the field, this article will bring a comprehensive view of the Internet of Things (IoT) within universities.

From the aged 2G era to now, short message is consistently playing an important role in both personal and commercial use. Along with the evolution of cellular systems, Rich Communication Services (RCS), which can provide rich-media messages and diverse interactions, has been proposed as an update of the obsolete text-based Short Message Services (SMS) to meet various requirements of modern applications. In China, the three major mobile network operators have jointly launched “5G Messaging Services (5GMS)”, a brand new RCS-based telecommunication business in the 5G era, to bring new user experience to individuals and valuable commercial opportunities to enterprises. Currently, most of the enterprise-level applications are delivered in the form of Software-as-a-Service (SaaS), and 5GMS can be utilized as a new entrance for various SaaS applications ideally thanks to its inherent advantages of nativeness, lightweight, interactivity, and security. In this paper, we design a 5GMS-to-SaaS middleware, which can deal with practical issues such as message transmission, format conversion, and account matching, to facilitate establishing connections between the unified front-end 5GMS and a large amount of various back-end SaaS applications. The proposed middleware can normalize and simplify the procedures of linking massive services, so as to promote development efficiency and accelerate business extension of various enterprise-level 5GMS-based SaaS applications.

Blockchain has received expanding interest from various domains. Institutions, enterprises, governments, and agencies are interested in Blockchain’s potential to augment their software systems. The unique requirements and characteristics of Blockchain platforms raise new challenges involving extensive enhancement to conventional software development processes to meet the needs of these domains. Software engineering approaches supporting Blockchain-oriented developments have been slow to materialize, despite proposals in the literature, and they have yet to be objectively analyzed. A critical appraisal of these innovations is crucial to identify their respective strengths and weaknesses. We present an analytical evaluation of several prominent Blockchain-oriented methods through a comprehensive, criteria-based evaluation framework. The results can be used for comparing, adapting, and developing a new generation of Blockchain-oriented software development processes and innovations.

Metaverse is the next generation gaming Internet, and virtual humans play an important role in Metaverse. The simultaneous representation of motions and emotions of virtual humans attracts more attention in academics and industry, which significantly improves user experience with the vivid continuous simulation of virtual humans. Different from existing work which only focuses on either the expression of facial expressions or body motions, this paper presents a novel and real-time virtual human prototyping system, which enables a simultaneous real-time expression of motions and emotions of virtual humans (short for SimuMan). SimuMan not only enables users to generate personalized virtual humans in the metaverse world, but also enables them to naturally and simultaneously present six facial expressions and ten limb motions, and continuously generate various facial expressions by setting parameters. We evaluate SimuMan objectively and subjectively to demonstrate its fidelity, naturalness, and real-time. The experimental results show that the SimuMan system is characterized by low latency, good interactivity, easy operation, good robustness, and wide application.

Power safety production has always been an important issue related to the national economy and people's livelihood in the energy system. For a long time, humans have relied only on manual inspection to monitor the transformer oil leakage hidden danger, through the camera multi angle continuous collection of data, and then used the long-term memory network and improved genetic algorithm combination, to solve the traditional statistical machine learning because the training sample is not enough to form artificial intelligence algorithm model of cold start problem. In this study, we used the architecture of cloud-edge collaboration to provide services. The complex large data model training is executed in the cloud, and then the model is written to the edge server for reasoning. At present, the system has completed the pilot operation in Beijing substation, and the operation effect is good. It can effectively identify all kinds of common oil leakage within 200 ms.

On the basis of introducing the metaverse and the digital economy and supporting the development trend of new technologies, the concept of service-to-service (S2S) ecosystem is defined. The metaverse is used to build a virtual digital world and link the physical world. This paper focuses on the connotation of the two words Meta and Verse of the metaverse, and proposes a management framework of metaverse resources (Non-fungible token, 3D space, experience, avatar, etc.) in scenarios that support interactions such as the management of work, life, transaction, and customers. Then the paper proposes a metaverse reference architecture (MRA) for systematically constructing metaverse solutions. Finally, this paper also looks forward to the future development trend of the metaverse.