Personal Technologies

We present a wearable input system which enables interaction through 3D handwriting recognition. Users can write text in the air as if they were using an imaginary blackboard. The handwriting gestures are captured wirelessly by motion sensors applying accelerometers and gyroscopes which are attached to the back of the hand. We propose a two-stage approach for spotting and recognition of handwriting gestures. The spotting stage uses a support vector machine to identify those data segments which contain handwriting. The recognition stage uses hidden Markov models (HMMs) to generate a text representation from the motion sensor data. Individual characters are modeled by HMMs and concatenated to word models. Our system can continuously recognize arbitrary sentences, based on a freely definable vocabulary. A statistical language model is used to enhance recognition performance and to restrict the search space. We show that continuous gesture recognition with inertial sensors is feasible for gesture vocabularies that are several orders of magnitude larger than traditional vocabularies for known systems. In a first experiment, we evaluate the spotting algorithm on a realistic data set including everyday activities. In a second experiment, we report the results from a nine-user experiment on handwritten sentence recognition. Finally, we evaluate the end-to-end system on a small but realistic data set.

Remote health monitoring (RHM) provides various benefits to older adults, but its use is still limited. Remote monitoring may help in avoiding emergencies and prolong users’ independence. To understand how to design systems which support older adults, we studied a large-scale remote health monitoring system. The system used fitness-grade smartwatches to monitor the vital signs of more than 2000 users constantly. To probe the lived experience of using RHM, we conducted an explorative interview study (N = 41) with operators, carers, and users of the RHM system. Our thematic analysis reveals that personalisation of care ecology is crucial for developing users’ confidence and trust in the system. We found that participation in RHM may catalyse positive changes in older adults’ lifestyles. Based on our findings, we formulate five recommendations for designing future health monitoring systems. Our work contributes to insights into the lived experience and stakeholder ecology of health monitoring systems.

In visual cryptographic scheme (VCS), each secret pixel is expanded to m subpixels in shadow images to encrypt a secret image. Apparently, we should put these m subpixels as a rectangle such that the blocks can be arranged compactly with each other. However, if the aspect ratio is viewed as important information of the secret image, the distortion occurs at the case that m is not a square. An aspect ratio invariant VCS (ARIVCS) was accordingly proposed to address the arrangement of subpixels. However, the existing ARIVCSs relied on adding dummy subpixels to the shadow images. In addition, their method needs a mapping pattern that reduces the number of dummy subpixels, but how to design a mapping pattern is a huge challenge. In this paper, we propose an easy solution to construct ARIVCS by image filtering and resizing.

We provide evidence of the feasibility and effectiveness of a middleware architecture for mobile devices (MoDs), which employs dense distributions of small computerized entities for providing fault-tolerant location-aware services. We do so by describing exemplary implementations based on radio frequency identification as an enabling technology. Firstly, we present prototypical implementations of the hardware abstraction layer and of selected core middleware services. The latter enable a MoD to store and retrieve data and position information in physical places in a fault-tolerant manner, and to identify places based on a location abstraction which is robust against failure of individual tags. Secondly, we investigate the feasibility of some higher-level services and applications by developing and evaluating prototypical systems for tracing and tracking, self-positioning, and collaborative map-making.

For more than a decade, it has been observed that due to the continuously increasing elderly population, medical expenditures for home health-care have rapidly increased every year. Health-care organizations are being encouraged to find more efficient methods of providing high-quality care with cost savings. Elderly people who live alone may fear being unable to obtain help if they are injured or ill. For many families, the fear of such an event can dominate decisions about living arrangements. During the last decades, this fear has generated an industry, marketing house automations, alarm-and-notification systems as well as assistive technologies and interventions that aim at easier and more efficient control of their living environment, as well as the provision of specialized home-care services. In this study, the most important aspects of the INHOME platform architecture, the home-care services provided together with the home-care applications developed such as health monitoring and household appliances monitoring and control, are discussed. The goal of the INHOME project is to provide ambient assistive living services to improve the quality of life of elderly people at home and is mainly targeted at ameliorating the ways of providing intelligent services for assisting independent living of elderly people at home.

Các mô hình nổi bật hình ảnh dựa trên các đặc trưng cấp thấp của hình ảnh gặp phải vấn đề về độ chính xác và khả năng mở rộng thấp, trong khi các mô hình nổi bật hình ảnh dựa trên mạng nơ-ron sâu có thể cải thiện hiệu suất dự đoán một cách hiệu quả, nhưng yêu cầu một lượng lớn dữ liệu huấn luyện, chẳng hạn như dữ liệu theo dõi ánh mắt, để đạt được kết quả tốt. Tuy nhiên, phương pháp theo dõi ánh mắt truyền thống bị giới hạn bởi chi phí thiết bị và thời gian cao, quy trình vận hành phức tạp, trải nghiệm người dùng kém, v.v. Do đó, bài báo này đã đề xuất một mô hình nổi bật hình ảnh dựa trên dữ liệu theo dõi ánh mắt từ crowdsourcing, được thu thập thông qua cách thức hồi tưởng ánh mắt với sự tự báo cáo từ những người làm việc trong đám đông. Tối ưu hóa tham số trên phương pháp crowdsourcing của chúng tôi đã được khám phá, và kết quả cho thấy độ chính xác của dữ liệu ánh mắt đạt 1° góc nhìn, cao hơn 3.6% so với các phương pháp crowdsourcing hiện có khác. Dựa trên điều này, chúng tôi đã thu thập một tập dữ liệu trang web của dữ liệu ánh mắt từ crowdsourcing và xây dựng một mô hình nổi bật hình ảnh dựa trên mạng nơ-ron tích chập hoàn toàn (FCN). Kết quả đánh giá cho thấy rằng sau khi được huấn luyện bằng dữ liệu ánh mắt từ crowdsourcing, mô hình đã hoạt động tốt hơn, chẳng hạn như độ chính xác dự đoán tăng 44.8%. Ngoài ra, mô hình của chúng tôi đã vượt trội hơn các mô hình nổi bật hình ảnh hiện có. Chúng tôi cũng đã áp dụng mô hình của mình để giúp các nhà thiết kế trang web đánh giá và điều chỉnh thiết kế hình ảnh của họ, và kết quả thử nghiệm cho thấy thiết kế đã chỉnh sửa nhận được đánh giá tốt hơn 8.2% so với thiết kế ban đầu.

Security and privacy for smart cities have attached much attention. As cloud computing can provide secure and reliable data storage and data sharing, it has become an important infrastructure for smart cities. However, in the cloud storage system, the cloud server cannot be fully trusted. Therefore, new solutions of access control of cloud data need to be proposed. In this paper, we construct a proxy-assisted access control scheme. We use the method of ciphertext policy attribute-based encryption (CP-ABE) to realize access control of cloud data for smart cities. Since major computation of decryption is outsourced to the cloud server, our scheme can achieve efficient user decryption. We also present an efficient user and attribute revocation method for our scheme which achieves both forward security and backward security. Finally, we give secure and performance analysis of our scheme, which show that our scheme is secure and highly efficient.

In Ubiquitous computing, small embedded sensor and computing nodes are the main enabling technologies. System programming for such small embedded systems is a challenging task involving various hardware components with different characteristics. This paper presents a file system which organizes all computational and sensory functionality of a sensor node as resources in a uniform name space. It further provides a lightweight and uniform access model for all these resources. This mechanism forms an abstraction from different hardware, makes functions re-useable and simplifies the development on such systems. With ParticleFS a concrete file system implementation on a sensor node platform is shown. Application cases demonstrate sensor logging, an interactive shell, executables, a pipe mechanism and remote access capabilities of the ParticleFS.