Catálogo de publicaciones - libros

Most commercialized speech recognition systems that have a large capacity and high recognition rates are a type of speaker dependent isolated word recognition systems. In order to extend the scope of recognition, it is necessary to increase the number of words that are to be searched. However, it shows a problem that exhibits a decrease in the system performance according to the increase in the number of words. This paper defines the context information that affects speech recognition in a ubiquitous environment to solve such a problem and designs a new speech recognition system that demonstrates better performances than the existing system by establishing a word model domain of a speech recognition system.

The distributed object group framework (DOGF) enables easier integration of distributed objects to healthcare home applications. This paper describes a healthcare information management system for supporting healthcare home services based on DOGF. The system architecture consists of healthcare database management tool, framework, sensor manager and physical elements such as sensors, devices and appliances. The focus of this paper is on the design and construction of a database system in a framework to support home healthcare services. Healthcare information is organized using database schemes based on the specific types of data collected from various typed-sensors. The database constructed from this information for the purpose of home healthcare services is divided into the base information that uses the real schemes and the context based information that uses the view schemes. To verify the practical use of the healthcare information management system proposed in this paper, we created a prototype healthcare home monitoring service using information, emergency call, and home appliance control. The result of the experimental evaluation shows the comparison of the execution service time of the base information and the context based information in our simulated scenario of a home healthcare application.

Smart Home System (SHS) is one of popular applications in ubiquitous computing, which provides convenient services for a user with user-friendly intelligent system interfaces. Among them, voice recognition is a popular interface. However, voice command statements given by users are often too unclear and incomplete for the devices in SHS to understand the original user intention. So, the devices become complicated and have no idea about whether to work or not. Therefore, we should make sure the proximate selection for the devices which will be eventually targeted and operated following user intention. In this paper, we propose an effective method to make a decision in electing a promising target device among candidates by taking advantage of complementary context feeding around user environment in SHS even with initial incomplete interface information. The proposed method is based on Bayes theorem using the way of empirical statistic inference.

We propose the reference implementation with software rendering of OpenVG for the scalable vector graphic hardware acceleration, which the Khronos group standardizes. We present the design scheme that enables EGL and OpenVG to be ported easily in an embedded environment. Moreover, we describe the background of selection of an algorithm, and the mathematical function adopted for the performance improvement, and we propose the optimum rendering method. We present displaying of vector image on a screen through the OpenVG implemented using software rendering method. And, we present the test result of the CTS which is compatibility test tool. And we show the performance comparison against the Hybrid corp.’s reference implementation.

Even without possible interference from external radio sources, the problem of guaranteeing quality-of-service (QoS) routing in mobile ad hoc networks (MANETs) is a difficult problem. Difficulties arise because node mobility can cause frequent network topology changes, communication channels can have high error rates, the jitter rate is high and several different applications can be sharing the use of the communication medium. This paper addresses these issues and proposes a new QoS routing protocol for MANETs that combines aspects of a MAC protocol and a path reservation protocol. The proposed QoS routing protocol can be implemented based on any routing method that supports multiple paths (e.g., DSDV, TORA, PDR). The QoS performance of the proposed protocol is verified with simulations conducted using NS-2.

To determine schedulability of priority-driven periodic tasksets on multi-processor systems, it is necessary to rely on utilization bound tests that are safe but pessimistic, since there is no known method for exact schedulability analysis for multi-processor systems analogous to the response time analysis algorithm for single-processor systems. In this paper, we use model-checking to provide a technique for exact multiprocessor scheduability analysis by modeling the real-time multi-tasking system with Timed Automata (TA), and transforming the schedulability analysis problem into the reachability checking problem of the TA model.

The hardware infrastructure that provides the support of ubiquitous embedded computing may be shared by different applications. Many of those applications have real-time requirements, where events from the environment require the reaction of the computing system. The meeting of deadlines is hindered by the fast system dynamics. At the same time, the embedded system must deal with overload situations. In this paper we assume that an embedded application receives aperiodic requests with soft deadlines. Other unknown applications are executed simultaneously. The goal of this paper is to discuss algorithms to estimate the probability of a deadline to be met. The prediction of a deadline miss at the request arrival allows actions for damage control.

The popularity of mobile and multimedia applications made real-time support a mandatory feature for embedded operating systems. However, the current situation is that the overall performance is significantly degraded due to the real-time support. This paper suggests a novel scheme to minimize the performance degradation in embedded operating systems with real-time support. Especially, we propose transparent and selective real-time interrupt services which transparently monitor the system and postpone interrupt handling that are not relevant to real-time tasks. The proposed scheme was implemented on the Linux 2.6 kernel and the experimental results show that our scheme improves the throughput by up to 86% for Hackbench benchmark while providing almost the same scheduling latency compared to the previous work.

This research studies energy-aware managemen strategies in ubiquitouse home network environment. Always-on connectivity is essential to establish home networks, and it induces more energy consumption. In this work, a methodology is prosposed to achieve energy saving in overalls without harming the quality of services. Energy management technology used in wireless sensor network is expanded to information appliances, usage patterns and dependecy tree are applieds to design the management system, and the effect in energy saving is analyzed.

With the significant driving force from the application domains, modern embedded systems are designed over heterogeneous multi-core SoC platforms. When more and more functions are integrated into one system, the designs of embedded systems have become more and more complicated. In particular, most of embedded multimedia applications are data intensive. Performance bottleneck are often caused by inappropriate bus architecture design within the system. In this paper, we present the algorithms for bus architecture optimization in MFASE. The algorithm takes the workloads in the system and their timing behavior requirements into account. The goal is to minimize the number of buses in the system without violating timing requirements. We prove that the minimzation problem is NP-hard and develop a heuristic algorithm. We evaluate the algorithm with extensive simulations. The performance results show that the algorithm reduce up to 80% of the bus cost and performs as well as optimal exponential algorithm does.