Search

In irregular repetition slotted ALOHA (IR-SA) systems, a population of devices transmits their packets to an access point (AP) within a frame of slots. The AP decodes these packets by iterative processing between intra- and inter-slot successive interference cancellations. The average normalized offered traffic, as a performance metric, represents the number of packets transmitted per slot when the packet loss rate approaches zero. Such asymptotic types of traffic as the belief propagation (BP) threshold, the maximum a posteriori (MAP) threshold, and the converse bound of IR-SA systems have been analyzed over various channel models. However, over fading channels, the MAP threshold and the converse bound have not yet been investigated. This paper derives an MAP threshold and a converse bound of the systems over Rayleigh block fading channels. The derivations are based on two extrinsic information transfer (EXIT) curves, which are associated with two iterative density evolution equations to analyze the BP threshold of the IR-SA systems. First, since an open decoding tunnel exists in an EXIT chart, the sum of the two areas below two EXIT curves is smaller than the area of the entire domain. This provides the traffic’s converse bound, which is tight. Second, a coincidence of the BP EXIT and MAP EXIT curves makes it possible to derive the traffic’s MAP threshold. Third, a density evolution for a spatially-coupled scheme is formulated and gives a BP decoding threshold of the traffic. Numerical results show that the spatially-coupled scheme achieves a threshold saturation effect where the BP threshold approaches the MAP threshold.

In this work, a prescribed-time second-order sliding mode (SOSM) controller is constructed to address a class of nonlinear systems in the presence of asymmetric output constraints. To handle the asymmetric constraints, a frequently-used asymmetric barrier Lyapunov function is employed. Meanwhile, a time-varying scaling function is introduced to ensure the prescribed-time convergence of the closed-loop sliding mode system. On this basis, a prescribed-time SOSM controller is constructed by inserting the asymmetric BLF and time-varying scaling function into a modified power integrator technique. The rigorous theoretical analysis indicates that under the proposed prescribed-time SOSM control scheme, the overall system is prescribed-time stable without violating the asymmetric constraints. The distinguishing feature is that the convergence time of the overall system can be specified beforehand and does not depend on any system initial conditions. Finally, an academic simulation example is provided to demonstrate the effectiveness of the presented prescribed-time control scheme.

In this paper, the issue of security sliding mode control (SMC) is addressed for networked nonlinear singular systems with semi-Markov switching parameters via the Takagi-Sugeno (T-S) fuzzy strategy. The goal is to guarantee a good steady performance of dynamical systems under the framework of deception attacks. First, the common sliding surface is proposed to avoid the potential instability caused by repeated jumps of sliding surface. Then, based on the partly unknown transition rate, sufficient conditions are constructed to realize the stochastic admissibility for the corresponding system. Moreover, a fuzzy security SMC law is designed to drive the state trajectory onto the specified sliding region. Finally, a practical example is shown to validate the proposed algorithm.

The RGB-D-based human action recognition is gaining increasing attention because the different modalities can provide complementary information. However, the recognition performance is still not satisfactory due to the limited ability to learn spatial-temporal feature and insufficient inter-model interaction. In this paper, we propose a novel approach for RGB-D human action recognition by aggregating spatial-temporal information and implementing cross-modality interactive learning. Firstly, a spatial-temporal information aggregation module (STIAM) is proposed to utilizes sample convolutional neural networks (CNNs) to aggregate the spatial-temporal information in entire RGB-D sequence into lightweight representations efficiently. This allows the model to extract richer spatial-temporal features with limited extra memory and computational cost. Secondly, a cross-modality interactive module (CMIM) is proposed to fully fuse the multi-modal complementary information. Moreover, a multi-modal interactive network (MMINet) is constructed for RGB-D-based action recognition by embeding the above two modules into the two-stream CNNs. In order to verify the universality of our approach, two backbones are deployed in the two-stream architecture, successively. Ablation experiments demonstrate that the proposed STIAM can bring significant improvement in recognizing actions. CMIM can further play the advantages of complementary features of multiple modalities. Extensive experiments on NTU RGB+D 60, NTU RGB+D 120 and PKU-MMD datasets proved the effectiveness of the proposed approach. The proposed approach achieves an accuracy of 94.3% and 96.5% for cross-subject and cross-view on NTU RGB+D 60, 91.7% and 92.6% for cross-subject and cross-setup on NTU RGB+D 120, 93.6% and 94.2% for cross-subject and cross-view on PKU-MMD datasets, which are the state-of-the-art performance. Further analysis denotes that our approach has advantages in recognizing subtle actions.

For a given software enhancement report, identifying its possible approval status could help software developers by suggesting feature enhancements to compete in the software industry. An automatic solution for the approval prediction of enhancements could assist all the participants in resolving enhancements. The key challenges are the preprocessing of noisy textual information and the state-of-the-art feature models to combine the syntactical and semantic word information available in the given text. To this end, we propose a deep learning based approach for the approval prediction of enhancement reports that incorporates the users’ sentiments involved in the text. First, we preprocess the textual information of all enhancement reports to avoid noise. Second, we compute the sentiment of each enhancement report using Senti4SD. Third, we combine the bag-of-words (BOW) representation and traditional word2vec based representation to learn the novel deep representation (a recurrent neural network (RNN) with attention based representation) of preprocessed text. Using an attention mechanism enables the model to remember the context over a long sequence of words in an enhancement report. Fourth, based on sentiment and deep representation, we train a deep learning based classifier for the approval prediction of enhancement reports. Finally, we reuse the 40, 000 enhancement reports from 10 real software applications to evaluate the proposed approach. The cross-application evaluation suggests that the proposed approach is accurate and outperforms the state-of-the-art. The results of the proposed approach improve the precision from 86.52% to 90.56%, recall from 66.45% to 80.10%, and f-measure from 78.12% to 85.01%.

The existing relation extraction approaches select the relevant partial dependency structures and exhibit the limitation associated with long-distance dependencies. Moreover, due to the repeated use of the irrelevant redundant information and the lack of consideration of the key semantic details, the extraction of relations is relatively complex when the entities overlap. To address this limitation and effectively exploit the relevant information while ignoring irrelevant information, this paper proposes a simple but effective multistream semantics-guided dynamic aggregation graph convolution network (SG-DAGCN) to realize the extraction of overlapping relations. The proposed model constructs the entity relation graphs by enumerating the possible candidates and external auxiliary information and adaptively manages the relevant substructure. Subsequently, this framework models the relational graphs between the entities through a dynamic aggregation graph convolution module and gradually produces the discriminative embedded features and a refined graph through the dynamic aggregation of nodes. The proposed approach can effectively leverage the rich multiscale structural information and capture the long-distance dependencies between overlapping entities in long sentences. The results of the experiments conducted on two typical benchmark datasets show that the proposed model can achieve a high level of performance and outperform other state-of-the-art methods in both qualitative and quantitative aspects.

With the rapid development and convenient service of online car-hailing, it has gradually become the preferred choice for people to travel. Accurate forecasting of car-hailing trip demand not only enables the drivers and companies to dispatch the vehicles and increase the mileage utilization, but also reduces the passengers' waiting-time. The rebalance of spatiotemporal demand and supply could mitigate traffic congestion, reduce traffic emission, and guide people's travel patterns. This study aimed to develop a short-term demand forecasting model for car-hailing services using stacking ensemble learning approach. The spatial-temporal characteristics of online car-hailing demand were analyzed and extracted through data analysis. The region-level spatial characteristics, time features, and weather conditions were added into the forecasting model. Then the stacking ensemble learning model was developed to predict the car-hailing demand at region-level for different time intervals, including 10 min, 15 min, and 30 min. The validation results suggested that the proposed stacking ensemble learning model has reasonable good prediction accuracy for different time intervals. The comparison results show that the short-term demand forecasting model based on stacking ensemble learning is better than single LSTM, SVR, lightGBM and Random Forest models. MAE and RMSE increased by 6.0% and 5.2% respectively at 30 min time interval, which further verifies the effectiveness and feasibility of the proposed model.

This work focuses on the issue of static output feedback control for Takagi-Sugeno (T-S) fuzzy systems in the discrete-time domain. Both the fading channel and actuator faults are determined by a set of stochastic variables. Specifically, an actuator fault is described by a nonhomogeneous Markov chain, and fading channels are characterized by the l-order Rician fading model. Using the Lyapunov-Krasovskii function, sufficient conditions are established, and controller gains are developed. Finally, the practical mass-spring-damping model is expressed to verify the practicability of the theoretical results.

This paper is concerned with the stability criteria of discrete neural networks with two additive input time-varying delays. By using the time delay division and a new summation inequality, a less conservative criterion is derived. Moreover, to compare the obtained criterion more directly with the existing results, a corollary is proposed accordingly. Finally, some numerical examples are presented to demonstrate that the obtained criteria are less conservative.

Few-shot learning aims to recognize novel categories from just a few labeled instances. Existing metric learning-based approaches perform classifications by nearest neighbor search in the embedding space. The embedding function is a deep neural network and usually shared by all novel categories. However, these brute approaches lack a fast adaptation mechanism like meta-learning when dealing with novel categories. To tackle this, we present a novel instance-level embedding adaptation mechanism, aiming at rapidly adapting embedding deep features to improve their generalization ability in recognizing novel categories. To this end, we design an Attention Adaptation Module to pull a query instance and its corresponding class center as close as possible. Note that, each query instance is pulled closer to its corresponding class center before performing nearest neighbor classifications. This instance-level reduction of intra-class distance increases the probability of correct classifications, and thus improves the generalization ability to embed deep features and promoting the performance. The extensive experiments are conducted on two benchmark datasets: miniImageNet and CUB. Our approach yields very promising results on both datasets. In addition, in a realistic cross-domain evaluation setting, our method also achieves the-state-of-the-art performance.

In recent years, an increasing number of university students are found to be at high risk of depression. Through a large scale depression screening, this paper finds that around 6.5% of the university postgraduate students in China experience depression. We then investigate whether the gait patterns of these individuals have already changed as depression is suggested to associate with gait abnormality. Significant differences are found in several spatiotemporal, kinematic and postural gait parameters such as walking speed, stride length, head movement, vertical head posture, arm swing, and body sway, between the depressed and non-depressed groups. Applying these features to classifiers with different machine learning algorithms, we examine whether natural gait analysis may serve as a convenient and objective tool to assist in depression recognition. The results show that when using a random forest classifier, the two groups can be classified automatically with a maximum accuracy of 91.58%. Furthermore, a reasonable accuracy can already be achieved by using parameters from the upper body alone, indicating that upper body postures and movements can effectively contribute to depression analysis.

Mobile cyber-physical systems (CPS) that take the advantages and extend the application domains of CPS have become increasingly popular in recent years. For example, mobile CPS could be a kind of foundational techniques to support the development of vehicular networking systems, thereby improving security and privacy of users in the dynamic environments of vehicular networks. In this paper, we first distinguish mobile CPS from traditional CPS. Then, we introduce their three emerging application areas, i.e., vehicular networking systems, healthcare systems, and mobile education. After that, we discuss four main research challenges of mobile CPS regarding security, energy consumption, mobile dynamic environment, and system stability. Also, we consider the corresponding techniques, which may address these challenges, and analyze the inter-relations among them. Finally, we outline the possible research directions and applications of mobile CPS in the future.

With rapid developments in cloud computing, artificial intelligence, and robotic systems, ever more complex tasks, such as space and ocean exploration, are being implemented by intelligent robots. Here, we propose an underwater image enhancement scheme for robotic visual systems. The proposed algorithm and its implementation enhances and outputs an image captured by an underwater robot in real time. In this scheme, pulse-coupled neural network (PCNN)-based image enhancement and color transfer algorithms are combined to enhance the underwater image. To avoid color imbalance in the underwater image and enhance details while suppressing noise, color correction is first carried out on the underwater image before converting it into the hue-saturation-intensity domain and enhancing it by PCNN. The enhanced result improves the color and contrast of the source image and enhances the details and edges of darker regions. Experiments are performed on real world data to demonstrate the effectiveness of the proposed scheme.

Human action recognition is one of the fundamental challenges in robotics systems. In this paper, we propose one lightweight action recognition architecture based on deep neural networks just using RGB data. The proposed architecture consists of convolution neural network (CNN), long short-term memory (LSTM) units, and temporal-wise attention model. First, the CNN is used to extract spatial features to distinguish objects from the background with both local and semantic characteristics. Second, two kinds of LSTM networks are performed on the spatial feature maps of different CNN layers (pooling layer and fully-connected layer) to extract temporal motion features. Then, one temporal-wise attention model is designed after the LSTM to learn which parts in which frames are more important. Lastly, a joint optimization module is designed to explore intrinsic relations between two kinds of LSTM features. Experimental results demonstrate the efficiency of the proposed method.

In multivariate time series clustering, the inter-similarity across distinct variates and the intra-similarity within each variate pose analytical challenges. Here, we propose a novel multivariate time series clustering method using multi-nonnegative matrix factorization (MNMF) in multi-relational networks. Specifically, a set of multivariate time series is transformed from the time-space domain into a multi-relational network in the topological domain. Then, the multi-relational network is factorized to identify time series clusters. The transformation from the time-space domain to the topological domain benefits from the ability of networks to characterize both the local and global relationships between the nodes, and MNMF incorporates inter-similarity across distinct variates into clustering. Furthermore, to trace the evolutionary trends of clusters, time series is transformed into a dynamic multi-relational network, thereby extending MNMF to dynamic MNMF. Extensive experiments illustrate the superiority of our approach compared with the current state-of-the-art algorithms.

Sparsely spread code division multiple access (SCDMA) is a non-orthogonal superposition coding scheme that allows concurrent communications between a base station and multiple users over a common channel. However, the detection performance of an SCDMA system is mainly determined by its signature matrix, which should be sparse to facilitate the belief propagation (BP) detection. On the other hand, to guarantee good maximum likelihood (ML) detection performance, the minimum Euclidean distance for the equivalent signal constellation after multi-user superposition should be maximized. In this paper, a code distance analysis is proposed for SCDMA systems with a finite number of users and spreading lengths. Based on this analysis, good signature matrices whose factor graphs have very few short cycles and possess large superposed signal constellation distances are designed. The proposed signature matrices have both good BP and ML detection performances. Moreover, their BP detection performances exactly converge to their ML detection performances with few iterations. It is worth pointing out that the proposed signature matrix design could be directly applied to the 5G non-orthogonal multiple access systems.

In today’s power electronic devices, power converters are widely applied in daily life, with increasing power demands. Consequently, the dynamic performance of power converters has become significant, as the compensators directly impact the system’s transient and dynamic response. This paper aims to design a type-III compensator for non-ideal buck converters to improve dynamic performance over a wide input voltage range. This paper introduces the non-ideal buck converter’s small signal model with state space averaging method, provides the guidelines for pole and zero positions of a type-III compensator using Matlab/Simulink Sisotool software, and presents simulation and experimental results to verify a design example and the performance and feasibility of the type-III compensator. Additionally, the transient and dynamic performance under different input voltages, load transients, and phase margins are compared with the K-factor and pole/zero compensation methods, providing a comprehensive reference when designing the compensator circuit for power converters. The results demonstrate that the designed type-III compensator effectively enhances stability and transient response, making it a valuable tool for engineers seeking to optimize power converter performance in various operational scenarios.

State-of-health (SOH) is a health index (HI) that directly reflects the performance degradation of lithium-ion batteries in engineering, but the SOH of Li-ion batteries is difficult to measure directly. In this paper, a novel data-driven method is proposed to estimate the SOH of Li-ion batteries accurately and explore the relationship-like mechanism. First, the features of the battery should be extracted from the performance data. Next, by using the evolution of genetic programming to reflect the change in SOH, a mathematical model describing the relationship between the features and the SOH is constructed based on the data. Additionally, it has strong randomness in the formula model, which can cover most of the structural space of SOH and features. An illustrative example is presented to evaluate the SOH of the two batches of Li-ion batteries from the NASA database using the proposed method. One batch of batteries was used for testing and comparison, and another was chosen to verify the test results. Through experimental comparison and verification, it is demonstrated that the proposed method is rather useful and accurate.