In Internet of Things (IoT) and industrial IoT (IIoT) systems, interconnected sensors continuously generate nonstationary multivariate time series (MTS), where inter-variable dependencies vary across operating regimes. Consequently, anomalies may manifest not only as point-wise deviations but also as abnormal changes in dependency patterns, especially lead–lag interactions induced by propagation delays and system responses. Reliably detecting such anomalies requires accurately modeling time-varying inter-variable dependencies. However, existing methods mainly focus on instantaneous correlations or lagagnostic dependency learning in the feature space, thereby over-looking lead–lag-aware dependency structures which are crucial for detecting anomalies in nonstationary environments. We propose CoLDM, an unsupervised reconstruction-based framework that learns lead-lag-aware inter-variable dependency representations for MTS anomaly detection. CoLDM first extracts statistical lead–lag cues and converts them into a soft structural prior, which is injected via Structural Preference Modulation to guide dependency learning. Moreover, CoLDM adopts a componentwise dual-branch architecture to capture complementary lead–lag dependency information, and introduces a component-adaptive latent fusion module that adaptively integrates the two representations across regimes, reducing redundant overlap and improving robustness. Experiments on real-world benchmarks demonstrate that CoLDM generalizes well and remains robust under nonstationary regimes.

With the rapid development of Internet of Things (IoT), an increasing volume of data is generated across diverse IoT devices. Within these data, an extremely limited subset may manifest notable deviations from the majority of data, such as network intrusion data in traffic monitoring devices. The identification of such anomalous data assumes considerable importance across diverse domains. In this article, we propose masked memory network, a semi-supervised anomaly detection method which can be applied to various IoT devices. Our approach leverages a masked memory module combined with a soft masking strategy to acquire discriminative patterns capable of distinguishing anomalies from normal data. We also devise a anomaly scoring strategy which can exploits the characteristics of attention weights between test samples and memory items to detect both known and unknown anomalies. Experimental results on various AD data sets collected by IoT devices demonstrate the effectiveness of our work in numerous IoT applications.

Accurate control of autonomous marine robots still poses challenges due to the complex dynamics of the environment. In this paper, we propose a Deep Reinforcement Learning (DRL) approach to train a controller for autonomous surface vessel (ASV) trajectory tracking and compare its performance with an advanced nonlinear model predictive controller (NMPC) in real environments. Taking into account environmental disturbances (e.g., wind, waves, and currents), noisy measurements, and non-ideal actuators presented in the physical ASV, several effective reward functions for DRL tracking control policies are carefully designed. The control policies were trained in a simulation environment with diverse tracking trajectories and disturbances. The performance of the DRL controller has been verified and compared with the NMPC in both simulations with model-based environmental disturbances and in natural waters. Simulations show that the DRL controller has 53.33% lower tracking error than that of NMPC. Experimental results further show that, compared to NMPC, the DRL controller has 35.51% lower tracking error, indicating that DRL controllers offer better disturbance rejection in river environments than NMPC.

Within the next several years, there will be a high level of autonomous technology that will be available for widespread use, which will reduce labor costs, increase safety, save energy, enable difficult unmanned tasks in harsh environments, and eliminate human error. Compared to software development for other autonomous vehicles, maritime software development, especially in aging but still functional fleets, is described as being in a very early and emerging phase. This presents great challenges and opportunities for researchers and engineers to develop maritime autonomous systems. Recent progress in sensor and communication technology has introduced the use of autonomous surface vehicles (ASVs) in applications such as coastline surveillance, oceanographic observation, multi-vehicle cooperation, and search and rescue missions. Advanced artificial intelligence technology, especially deep learning (DL) methods that conduct nonlinear mapping with self-learning representations, has brought the concept of full autonomy one step closer to reality. This article reviews existing work on the implementation of DL methods in fields related to ASV. First, the scope of this work is described after reviewing surveys on ASV developments and technologies, which draws attention to the research gap between DL and maritime operations. Then, DL-based navigation, guidance, control (NGC) systems and cooperative operations are presented. Finally, this survey is completed by highlighting current challenges and future research directions.

Erasure-coded redundancy is a fault-tolerant method with low-cost storage overhead. It only stores data fragments and parity fragments rather than full data across the cluster. The write process of erasure-coded data can be asynchronous or synchronous. For synchronous write process, data are encoded when written to servers. The common method doing the process needs to confirm that each coded-fragment of the data is stored in a different server to maintain the best fault tolerance. This method underperforms in terms of availability, and also fails to achieve good performance because any failure of servers will shortly disturb the write process. Some consensus protocols such as RS- Paxos and CRaft, which are based on Paxos and Raft, can solve above problems by providing fault-tolerant ability for systems. However, RS-Paxos cannot achieve the same liveness as Paxos. CRaft still adopts full data redundancy to keep the same liveness as Raft when there are not enough healthy servers. Therefore, to solve the availability problem during synchronous erasure-coded data write process, we present a novel protocol ECRaft based on Raft. It always uses erasure-coded redundancy when the ratio of erasure-coded data fragments to parity fragments is bigger than 1. It also can reach the same liveness as Raft. With state machine purge, storage redundancy can be reduced to the extent that typical erasure-coded storage systems can achieve. We build a key-value store based on ECRaft to evaluate it. In our experiments, compared with CRaft using complete-entry replication, ECRaft can save 63 % of storage, increase write throughput by 28.2 %, and reduce write latency by 19 %.

Segmentation from renal pathological images is a key step in automatic analyzing the renal histological characteristics. However, the performance of models varies significantly in different types of stained datasets due to the appearance variations. In this paper, we design a robust and flexible model for cross-stained segmentation. It is a novel multi-level deep adversarial network architecture that consists of three sub-networks: (i) a segmentation network; (ii) a pair of multi-level mirrored discriminators for guiding the segmentation network to extract domain-invariant features; (iii) a shape discriminator that is utilized to further identify the output of the segmentation network and the ground truth. Experimental results on glomeruli segmentation from renal biopsy images indicate that our network is able to improve segmentation performance on target type of stained images and use unlabeled data to achieve similar accuracy to labeled data. In addition, this method can be easily applied to other tasks.

Linking online identities of users among countless heterogeneous network services on the Internet can provide an explicit digital representation of users, which can benefit both research and industry. In recent years, user identity linkage (UIL) through the Internet has become an emerging task with great potential and many challenges. Existing works mainly focus on online social networks that consider inconsistent profiles, content, and networks as features or use sparse location-based data sets to link the online behaviors of a real person. To extend the UIL problem to a general scenario, we try to link the web-browsing behaviors of users, which can help to distinguish specific users from others, such as children or malicious users. More specifically, we propose a Siamese neural network (NN) architecture-based UIL (SAUIL) model that learns and compares the highest-level feature representation of input web-browsing behaviors with deep NNs. Although the number of matching and nonmatching pairs for the UIL problem is highly imbalanced, previous studies have not considered imbalanced UIL data sets. Therefore, we further address the imbalanced learning issue by proposing cost-sensitive SAUIL (C-SAUIL) model, which assumes higher costs for misclassifying the minority class. In the experiments, the proposed model is robust and exhibits a good performance on very large, real-world data sets collected from different regions with distinct characteristics.

In recent years, with the rapid development of Internet, more and more users shop and socialize online, which produces a large amount of traffic data. Analyzing users' online shopping behavior is important for merchants to improve profit. Social networks are widely used in studying the relationship among users. In this paper, we apply the concept of social networks to online shopping behavior, and present a novel network perspective on the interconnected nature of online interest and product, allowing us to capture the attribute of online products reflecting users' sociability. Sociability in this paper is not a traditional social connection but common interests(defined by produce visiting behavior) among users. First, we build an interconnected online interest-product network which includes online shopping based social network (OSSN), online product network and intermediate layer, and we define popular online products and non-popular online products according to the number of distinct visitors of one product. Then, we analyze some indicators about OSSN, and find that OSSN has similar characteristics with traditional social networks, such as small world feature and homophily. Finally, we analyze how online products reflect users' sociability. With the number of online products users have visited increasing, the number of neighbors with similar interests first presents a positive correlation, and then disappears, which implies that the number of users' neighbors depends on the attribute of online products rather than the number of online products users have browsed. We define a formula to measure the attribute of online products reflecting users' sociability, and find that there is a distinct difference among different categories of online products, and non-popular online products often have high value of attribute to reflect users' sociability and the value is low in popular online products.

This paper attempts to identify the requirement and the development of machine learning‐based mobile big data (MBD) analysis through discussing the insights of challenges in the mobile big data. Furthermore, it reviews the state‐of‐the‐art applications of data analysis in the area of MBD. Firstly, we introduce the development of MBD. Secondly, the frequently applied data analysis methods are reviewed. Three typical applications of MBD analysis, namely, wireless channel modeling, human online and offline behavior analysis, and speech recognition in the Internet of Vehicles, are introduced, respectively. Finally, we summarize the main challenges and future development directions of mobile big data analysis.

Due to geographic and social constraints, human mobility shows a high degree of temporal and spatial regularity. More recently, emerging data tagged with geographical information can be used to study human mobility patterns. People usually spend most of their time at a few key important locations, such as home and work places. And for users with a certain social connection-co-workers or co-life, they often stay at the same locations. In this paper, firstly, we use an algorithm to identify important locations. After that, we find that the similarity between the two trajectories is closely related to their proximity in the location-based social network, where users having the same important locations are connected. In order to embody social contacts in mobility, for hourly variations in mobility similarity, we apply unsupervised clustering method to identify four categories of social ties. Finally, we further propose the unsupervised method and supervised method to predict which new links will develop in a social location-based network. We believe our finding can contribute to urban planning especially in areas of functional zone, transportation infrastructure deployment and mobile network facilities development.

In the era of the Internet, people are active in multiple online services, and they usually have accounts on more than one online service. Each account is a virtual identity of the user. In order to trace individual's online behavior at any time and any places, linking virtual identities belonging to the same natural person across different online service domains is very important. Existing methods usually tackle this problem by estimating the profile content similarity between identities under two different online services. However, the profile contents in various online services are unreliable or misaligned, and the proposed methods are always limited to services in a specific domain. In this paper, we propose VISD (Virtual Identity linkage cross Service Domain), a novel probability-based model, to link virtual identities across online services in various service domains. It derives several significant attributes from users' online behaviors, such as IP address usage, various fingerprints of terminals, and leverages a supervised classification method to discover the relationship between two identities. By using real-world network traffic collected from a large province of southern China, we evaluate the VISD model and the linkage precision achieves 88.31%. The result demonstrates the effectiveness of our proposed model, which is helpful for social recommendations, information security and privacy protection.

In recent years, the explosion of mobile Internet traffic data not only brings a high commercial value, but also helps us get a better understanding of human behavior. The study of human mobility which aims at revealing the general laws of human movement is the basis of many social, economic, and technological phenomena. It has received wide attention. But at present, the researches on the relationship between user offline and online behavior is very limited, only stop at discovering that users' current locations impact their application access behavior. This paper attempts to explore the relationship between human offline mobility and online browsing tendency from a more general point of view. To achieve this goal, we need to address several challenges including the effect of data size on the credibility of experimental results, urban functional regions identification, the modeling of user offline mobility and online browsing tendency. In this paper, our dataset consists of 7 days real mobile Internet traffic in a northern city of China, which covers 181873 users' traffic logs. In addition, we improve the existing urban functional regions identification method, and propose a new spatio-temporal-based user mobility model - user mobility image. Finally we analyze the correlation between user offline behavior and online browsing tendency. The result of this work paves the way for the identification of urban functional regions based on base station patterns. Compared with spatial-based mobility model, proposed user mobility image is more effective in illustrating user offline mobility, and shows high accuracy when predicting online browsing tendency.

Due to the accelerating rate of development of Mobile Internet, the role of mobile phone becomes increasingly vital in people's life. Through the data generated by mobile phones, users' mobility characteristic and online interests can be roundly mined. By connecting human mobility and online behavior, many achievements have been proposed and then applied in many services to improve the quality of people's life. In this paper, by using the cellular network data collected from certain urban city in northeast China over a month, we first extract users' important locations. Based on this, we apply the topic model to extract mobility topics of users and explore users' application usage. We observe that different user mobility properties accompanied with different online behavior tendencies of users. Besides, we extract cross-town events based on users' important locations to explore the macroscopical organization of the city. The results of our research provide a comprehensive understanding of user behavior in Mobile Internet and city structure, which are valuable for the practitioners in the fields of Location-Based Services (LBS) and urban planning.

The developments of Information and Communication Technology (ICT) and Internet of Things (IoT) are being used to enhance quality, performance and interactivity of urban services. Benefited from the widespread adoption of mobile devices, we can collect amount of mobile data for user mobility analysis. Mining hidden information from users' mobile data is important for builders of smart city to provide better location-based service. This paper focuses on two classical domain-independent prediction models and one improved Markov model that are capable of estimating the next location. By using 27-day-long traffic data of mobile network, we extract trajectories of 4914 individuals for experiments. We find that the original Markov algorithm has a better performance in resource consumption than LZ family algorithms, but its prediction accuracy is lower than prediction accuracy of LeZi Update and Active LeZi algorithm. In order to improve the prediction accuracy of Markov and overcome drawbacks of traditional prediction algorithms, we present a new method based on Markov, which considers both temporal and spatial factors. Extensive experiments demonstrate our improved method has a better performance in location prediction. In addition, we further study the relationship between prediction accuracy and trajectory's regularity, to identify the most suitable prediction algorithm for a trajectory.

Data collected from the mobile Internet have the potential knowledge to provide important human mobility patterns. Understanding human mobility patterns is important to many location-based services, and could be used to predict users' behavior. In this paper, we concentrate on the issue of discovering human mobility patterns on both global and individual levels based on hotspots. We study the human mobility trajectories during 22 days for 3474 individuals collected at the core of a metropolitan Long Term Evolution (LTE) network in China. We employ a parameter-free method to detect hotspots, and demonstrate the effectiveness of our mobility pattern discovery algorithm by using the hotspots identified on both global and individual levels. We analyze the occurrence time distribution of these patterns and find that the global mobility patterns have higher occurrence probability in the morning, which indicates that people in a city tend to share the common commuting routes. For individual mobility patterns, there exists a strong spatiotemporal correlation property, implying that the individual mobility patterns have their own typical occurrence time depending on the pattern's context.

Recently, mobile Internet gained a strong momentum of development, which has led to increasing demand on mobile network traffic characterization and modeling. A good model of mobile network traffic can be used to make accurate prediction regarding various performance metrics. Based on the network trace collected from network backbone, our paper studies mobile network traffic characteristics in terms of the flow arrival numbers and flow connection duration. Basically, we employ the Poisson regression from Generalized Linear Model with time window clustering so as to approximate a time-dependent Poisson Process to the flow arrival process. Our analytical results demonstrate the accuracy of the adopted approach. In addition, through approximating the Phase Type distribution to the heavy-tailed distribution, our paper also models the flow connection duration. The obtained results can help us get a comprehensive understanding of the network performance, in accordance with which the resource usage may be optimized, e.g., we can expand network bandwidth or increase the buffer size when the network arrival is high.

A novel Bayesian matrix factorization method for bounded support data is presented. Each entry in the observation matrix is assumed to be beta distributed. As the beta distribution has two parameters, two parameter matrices can be obtained, which matrices contain only nonnegative values. In order to provide low-rank matrix factorization, the nonnegative matrix factorization (NMF) technique is applied. Furthermore, each entry in the factorized matrices, i.e., the basis and excitation matrices, is assigned with gamma prior. Therefore, we name this method as beta-gamma NMF (BG-NMF). Due to the integral expression of the gamma function, estimation of the posterior distribution in the BG-NMF model can not be presented by an analytically tractable solution. With the variational inference framework and the relative convexity property of the log-inverse-beta function, we propose a new lower-bound to approximate the objective function. With this new lower-bound, we derive an analytically tractable solution to approximately calculate the posterior distributions. Each of the approximated posterior distributions is also gamma distributed, which retains the conjugacy of the Bayesian estimation. In addition, a sparse BG-NMF can be obtained by including a sparseness constraint to the gamma prior. Evaluations with synthetic data and real life data demonstrate the good performance of the proposed method.