In this paper we investigate the reliability of general type shared protection systems i.e. M for N (M:N) that can typically be applied to various telecommunication network devices. We focus on the reliability that is perceived by an end user of one of N units. We assume that any failed unit is instantly replaced by one of the M units (if available). We describe the effectiveness of such a protection system in a quantitative manner. The mathematical analysis gives the closed-form solution of the availability, the recursive computing algorithm of the MTTFF (Mean Time to First Failure) and the MTTF (Mean Time to Failure) perceived by an arbitrary end user. We also show that, under a certain condition, the probability distribution of TTFF (Time to First Failure) can be approximated by a simple exponential distribution. The analysis provides useful information for the analysis and the design of not only the telecommunication network devices but also other general shared protection systems that are subject to service level agreements (SLA) involving user-perceived reliability measures.

Context-aware systems detect user's physical and social contexts based on sensor networks, and provide services that adapt to the user accordingly. Representing, detecting, and managing the contexts are important issues in context-aware systems. Composition of contexts is a useful method for these works, since it can detect a context by automatically composing small pieces of information to discover service. Danger-aware services are a kind of context-aware services which need description of relations between a user and his/her surrounding objects and between users. However when applying the existing composition methods to danger-aware services, they show the following shortcomings that (1) they have not provided an explicit method for representing composition of multi-user' contexts, (2) there is no flexible reasoning mechanism based on similarity of contexts, so that they can just provide services exactly following the predefined context reasoning rules. Therefore, in this paper, we propose a two-stage composition method based on context similarity to solve the above problems. The first stage is composition of the useful information to represent the context for a single user. The second stage is composition of multi-users' contexts to provide services by considering the relation of users. Finally the danger degree of the detected context is computed by using context similarity between the detected context and the predefined context. Context is dynamically represented based on two-stage composition rules and a Situation theory based Ontology, which combines the advantages of Ontology and Situation theory. We implement the system in an indoor ubiquitous environment, and evaluate the system through two experiments with the support of subjects. The experiment results show the method is effective, and the accuracy of danger detection is acceptable to a danger-aware system.

Automatic recognition of finger gestures can be used for promotion of life quality. For example, a senior citizen can control the home appliance, call for help in emergency, or even communicate with others through simple finger gestures. Here, we focus on one-stroke finger gesture, which are intuitive to be remembered and performed. In this paper, we proposed and evaluated an accelerometer-based method for detecting the predefined one-stroke finger gestures from the data collected using a MEMS 3D accelerometer worn on the index finger. As alternative to the optoelectronic, sonic and ultrasonic approaches, the accelerometer-based method is featured as self-contained, cost-effective, and can be used in noisy or private space. A compact wireless sensing mote integrated with the accelerometer, called MagicRing, is developed to be worn on the finger for real data collection. A general definition on one-stroke gesture is given out, and 12 kinds of one-stroke finger gestures are selected from human daily activities. A set of features is extracted among the candidate feature set including both traditional features like standard deviation, energy, entropy, and frequency of acceleration and a new type of feature called relative feature. Both subject-independent and subject-dependent experiment methods were evaluated on three kinds of representative classifiers. In the subject-independent experiment among 20 subjects, the decision tree classifier shows the best performance recognizing the finger gestures with an average accuracy rate for 86.92 %. In the subject-dependent experiment, the nearest neighbor classifier got the highest accuracy rate for 97.55 %.

In Still-to-Video (S2V) face recognition, only a few high resolution images are registered for each subject, while the probe is video clips of complex variations. As faces present distinct characteristics under different scenarios, recognition in the original space is obviously inefficient. Thus, in this paper, we propose a novel discriminant analysis method to learn separate mappings for different scenario patterns (still, video), and further pursue a common discriminant space based on these mappings. Concretely, by modeling each video as a manifold and each image as point data, we form the scenario-oriented mapping learning as a Point-Manifold Discriminant Analysis (PMDA) framework. The learning objective is formulated by incorporating the intra-class compactness and inter-class separability for good discrimination. Experiments on the COX-S2V dataset demonstrate the effectiveness of the proposed method.

Arterial spin labeling (ASL) is a non-invasive magnetic resonance imaging (MRI) method that can provide direct and quantitative measurements of cerebral blood flow (CBF) of scanned patients. ASL can be utilized as an imaging modality to detect Alzheimer's disease (AD), as brain atrophy of AD patients can be revealed by low CBF values in certain brain regions. However, partial volume effects (PVE), which is mainly caused by signal cross-contamination due to voxel heterogeneity and limited spatial resolution of ASL images, often prevents CBF in ASL from being precisely measured. In this study, a novel PVE correction method is proposed based on pixel-wise voxels in ASL images; it can well handle with the existing problems of blurring and loss of brain details in conventional PVE correction methods. Dozens of comparison experiments and statistical analysis also suggest that the proposed method is superior to other PVE correction methods in AD diagnosis based on real patients data.

In communication protocols, the behaviour of a protocol entity is related to the behaviour of another protocol entity as they communicate under sets of communication rules (protocols). Thus, it is desirable to concentrate on the design of one protocol entity and generate the corresponding protocol entity automatically. Furthermore, it is desirable that the protocol is formal, precise and unambiguous that is, it is described using FDTs (Formal Description Techniques). In this paper, we propose a protocol synthesis algorithm in which, from a LOTOS specification of a single given entity, LOTOS specification of the corresponding peer entity is generated automatically. Unlike previous works, where FSMs (Finite State Machines) were used to synthesize protocols, we use LOTOS, which is one of FDTs developed by ISO, in our proposed synthesis algorithm. We prove that the generated protocol is logical errors free, collectively represented as deadlock free, if the given entity is in certain forms which are natural in the context of connunication protocols.

A fast algorithm to speed up the search process of vector quantization encoding is presented. Using the sum and the partial norms of a vector, some eliminating inequalities are constructeded. First the inequality based on the sum is used for determining the bounds of searching candidate codeword. Then, using an inequality based on subvector norm and another inequality combining the partial distance with subvector norm, more unnecessary codewords are eliminated without the full distance calculation. The proposed algorithm can reject a lot of codewords, while introducing no extra distortion compared to the conventional full search algorithm. Experimental results show that the proposed algorithm outperforms the existing state-of-the-art search algorithms in reducing the computational complexity and the number of distortion calculation.

A simple and successful design of initial codebook of vector quantization (VQ) is presented. For existing initial codebook algorithms, such as random method, the initial codebook is strongly influenced by selection of initial codewords and difficult to match with the features of the training vectors. In the proposed method, training vectors are sorted according to the norm of training vectors. Then, the ordered vectors are partitioned into N groups where N is the size of codebook. The initial codewords are obtained from calculating the centroid of each group. This initializtion method has a robust performance and can be combined with the VQ algorithm to further improve the quality of codebook.

Context awareness is viewed as one of the most important goals in the pervasive computing paradigm. As one kind of context awareness, danger awareness describes and detects dangerous situations around a user, and provides services such as warning to protect the user from dangers. One important problem arising in danger-aware systems is that the description/definition of dangerous situations becomes more and more complex, since many factors have to be considered in such description, which brings a big burden to the developers/users and thereby reduces the reliability of the system. It is necessary to develop a flexible reasoning method, which can ease the description/definition of dangerous situations by reasoning dangers using limited specified/predefined contexts/rules, and increase system reliability by detecting unspecified dangerous situations. Some reasoning mechanisms based on context similarity were proposed to address the above problems. However, the current mechanisms are not so accurate in some cases, since the similarity is computed from only basic knowledge, e.g. nature property, such as material, size etc, and category information, i.e. they may cause false positive and false negative problems. To solve the above problems, in this paper we propose a new flexible and accurate method from feature point of view. Firstly, a new ontology explicitly integrating basic knowledge and danger feature is designed for computing similarity in danger-aware systems. Then a new method is proposed to compute object similarity from both basic knowledge and danger feature point of views when calculating context similarity. The method is implemented in an indoor ubiquitous test bed and evaluated through experiments. The experiment result shows that the accuracy of system can be effectively increased based on the comparison between system decision and estimation of human observers, comparing with the existing methods. And the burden of defining dangerous situations can be decreased by evaluating trade-off between the system's accuracy and burden of defining dangerous situations.

Image quality assessment (IQA) is an inherent problem in the field of image processing. Recently, deep learning-based image quality assessment has attracted increased attention, owing to its high prediction accuracy. In this paper, we propose a fully-blind and fast image quality predictor (FFIQP) using convolutional neural networks including two strategies. First, we propose a distortion clustering strategy based on the distribution function of intermediate-layer results in the convolutional neural network (CNN) to make IQA fully blind. Second, by analyzing the relationship between image saliency information and CNN prediction error, we utilize a pre-saliency map to skip the non-salient patches for IQA acceleration. Experimental results verify that our method can achieve the high accuracy (0.978) with subjective quality scores, outperforming existing IQA methods. Moreover, the proposed method is highly computationally appealing, achieving flexible complexity performance by assigning different thresholds in the saliency map.

We analyze the beat noise cancellation in two-dimensional optical code-division multiple-access (2-D OCDMA) systems using an optical hard-limiter (OHL) array. The Gaussian shape of optical pulse is assumed and the impact of pulse propagation is considered. We also take into account the receiver noise and multiple access interference (MAI) in the analysis. The numerical results show that, when OHL array is employed, the system performance is greatly improved compared with the cases without OHL array. Also, parameters needed for practical system design are comprehensively analyzed.

In this paper, a novel model of Gaussian pulse propagation in optical fiber is proposed to comprehensively analyze the impact of Group Velocity Dispersion (GVD) on the performance of two-dimensional wavelength hopping/time spreading optical code division multiple access (2-D WH/TS OCDMA) systems. In addition, many noise and interferences, including multiple access interference (MAI), optical beating interference (OBI), and receiver's noise are included in the analysis. Besides, we propose to use the heterodyne detection receiver so that the receiver's sensitivity can be improved. Analytical results show that, under the impact of GVD, the number of supportable users is extremely decreased and the maximum transmission length (i.e. the length at which BER 10-9 can be maintained) is remarkably shortened in the case of normal single mode fiber (ITU-T G.652) is used. The main factor that limits the system performance is time skewing. In addition, we show how the impact of GVD is relieved by dispersion-shifted fiber (ITU-T G.653). For example, a system with 321 Gbit/s users can achieve a maximum transmission length of 111 km when transmitted optical power per bit is -5 dBm.

With the progress of ubiquitous technology, ubiquitous learning presents new opportunities to learners. Situations of a learner can be grasped through analyzing the learner's actions collected by sensors, RF-IDs, or cameras in order to provide support at proper time, proper place, and proper situation. Training for acquiring skills and enhancing physical abilities through exercise and experience in the real world is an important domain in u-learning. A training program may last for several days and has one or more training units (exercises) for a day. A learner's performance in a unit is considered as short term state. The performance in a series of units may change with patterns: progress, plateau, and decline. Long term state in a series of units is accumulatively computed based on short term states. In a learning/training program, it is necessary to apply different support strategies to adapt to different states of the learner. Adaptation in learning support is significant, because a learner loses his/her interests easily without adaptation. Systems with the adaptive support usually provide stimulators to a learner, and a learner can have a great motivation in learning at beginning. However, when the stimulators reach some levels, the learner may lose his/her motivation, because the long term state of the learner changes dynamically, which means a progress state may change to a plateau state or a decline state. In different long term learning states, different types of stimulators are needed. However, the stimulators and advice provided by the existing systems are monotonic without changeable support strategies. We propose a mutual adaptive support. The mutual adaptation means each of the system and the learner has their own states. On one hand, the system tries to change its state to adapt to the learner's state for providing adaptive support. On the other hand, the learner can change its performance following the advice given based on the state of the system. We create a ubiquitous pet (u-pet) as a metaphor of our system. A u-pet is always with the learner and encourage the leaner to start training at proper time and to do training smoothly. The u-pet can perform actions with the learner in training, change its own attributes based on the learner's attributes, and adjust its own learning rate by a learning function. The u-pet grasps the state of the learner and adopts different training support strategies to the learner's training based on the learner's short and long term states.

This paper proposes to obtain high-level, domain-robust representations for cross-view face recognition. Specially, we introduce Convolutional Deep Belief Networks (CDBN) as the feature learning model, and an CDBN based interpolating path between the source and target views is built to model the correlation of cross-view data. The promising results outperform other state-of-the-art methods.

This paper introduces recent trends in video streaming and four methods proposed by the authors for video streaming. Video traffic dominates the Internet as seen in current trends, and new visual contents such as UHD and 360-degree movies are being delivered. MPEG-DASH has become popular for adaptive video streaming, and machine learning techniques are being introduced in several parts of video streaming. Along with these research trends, the authors also tried four methods: route navigation, throughput prediction, image quality assessment, and perceptual video streaming. These methods contribute to improving QoS/QoE performance and reducing power consumption and storage size.

High Efficiency Video Coding (HEVC/H.265) obtains 50% bit rate reduction than H.264/AVC standard with comparable quality at the cost of high computational complexity. Merge mode is one of the most important new features introduced in HEVC's inter prediction. Merge mode and traditional inter mode consume about 90% of the total encoding time. To address this high complexity, this paper utilizes the merge mode to accelerate inter prediction by four strategies. 1) A merge candidate decision is proposed by the sum of absolute transformed difference (SATD) cost. 2) An early merge termination is presented with more than 90% accuracy. 3) Due to the compensation effect of merge candidates, symmetric motion partition (SMP) mode is disabled for non-8×8 coding units (CUs). 4) A fast coding unit filtering strategy is proposed to reduce the number of CUs which need to be fine-processed. Experimental results demonstrate that our fast strategies can achieve 35.4%-58.7% time reduction with 0.68%-1.96% BD-rate increment in RA case. Compared with similar works, the proposed strategies are not only among the best performing in average-case complexity reduction, but also notably outperforming in the worst cases.

In this paper, we present an automatic implementation method by which executable communication programs in C can be generated from protocol specifications in LOTOS. The implementation method consists of two parts: 1) An implementation strategy and 2) a set of translation rules. The first part consists of the basic ideas on how to realize the primary mechanisms in LOTOS specifications. The second part formulates the implementation method by way of the translation rules based on the implementation strategy. The characteristics of our method can be summarized as follows: We formulate our implementation method by way of translation rules. These rules are defined topdown in the form of syntax-directed translation function. The mechanism for controlling concurrency and communication among the user processes corresponding to the processes in LOTOS specification is easily realized by using UNIX operating system functions. The translation rules have been implemented on the AS 3000 (SUN3) workstation. An application of this implementation method is demonstrated by a simplified token-ring-protocol.