Spatiotemporal data collected from real world often suffers from data missing, which causes difficulties for subsequent analysis and use. Most current spatiotemporal data imputation methods employ the spatiotemporal attention mechanism to capture the spatiotemporal features of the data. However, the spatiotemporal attention mechanism has numerous parameters and high computational complexity, which creates challenges for the model expansion and implementation. In addition, most of these methods simply stack spatial graph features and spatiotemporal features without sufficient feature fusion. To this end, the spatiotemporal feature fusion (STFF) model for the spatiotemporal data imputation is proposed in this paper. The fusion module of STFF can simultaneously capture and fuse temporal and spatial features with fewer parameters and lower computational complexity than the spatiotemporal attention mechanism. The differential module of STFF eliminates common features and amplifies differential features through feature differentiation, thereby addressing the over-smoothing problem of the fusion module. The feature compression and fusion module of STFF deeply fuses spatiotemporal features and spatial graph features based on the idea of the squeeze-and-excitation network. Experiments on 6 real spatiotemporal datasets demonstrate that STFF outperforms current spatiotemporal data imputation methods.

Wireless sensor networks (WSNs) have been widely applied to environmental monitoring and smart cities. Wireless sensors need to continuously collect and transfer information for long periods without human maintenance. Therefore, improving the efficiency of information transmission is a major challenge in the design of routing protocols for WSNs, which should enable wireless sensors to conserve and balance energy of wireless sensors, while ensuring high information transmission quality. To handle the limitations of current routing protocols and achieve efficiency information transmission, a distributed Q-learning based routing protocol is proposed in this paper. In the proposed protocol, five information transmission factors, which are the information transmission distance, direction and quality, as well as the residual energy and energy consumption of wireless sensors, are considered in the Q-value function to achieve high efficiency information transmission. In addition, a selective Q-value update strategy is proposed to enable our protocol to adapt to large-scale WSNs. Simulation results show that, compared with representative existing routing protocols (RLBR, UWSN, and LQEAR), the proposed protocol extends the network lifetime by about 17

With the rapid development of social networks, the speed and volume of information spread are increasing. However, the spread of misinformation in social networks may lead to public opinion deviation and have negative influence on the society. With the scale expansion of social networks, it is impossible to have a central controller to detect the misinformation and minimize its influence in social networks. To this end, this paper proposes a decentralized method to handle the misinformation in social networks, which enables users to cooperatively shift the attention of the target user from the misinformation, so as to reduce the influence of misinformation to the society. In our method, an intelligent agent based model is established to simulate users and their relationships in social networks. Then, intelligent agents can explore message sending paths to the target user in a decentralized manner. Finally, suitable messages and their sending paths are dynamically selected to efficiently and effectively shift the attention of the target user. The experiments on real social network datasets (i.e., Twitter, Wiki-Vote, Epinions) indicate the good performance of the proposed method in terms of the attention shift of the target user and minimizing the influence of misinformation on the society.

With the rapid development of geolocation technology, the volume of spatio-temporal trajectory data has surged. This data is widely used in fields such as geographic information systems and mobile computing, but its storage and query processing present significant challenges. Current methods of offline indexing are inefficient and cannot be updated in real-time. To address this issue, this paper proposes a concept of the online index that supports real-time storage and indexing of trajectory data and significantly reduces indexing time and storage space requirements. Based on this concept, two vector-based online trajectory indexing methods are proposed in this paper. The first is an online trajectory indexing method based on vector extraction (VBIndex), which offers the advantages of high efficiency and less storage space. The second is an online trajectory indexing method based on road-network matching (RAIndex), which further improves the vector-based indexing efficiency when road network involved. Through experiments with real datasets, the proposed algorithms were evaluated, confirming their superiority in terms of indexing construction time and storage space. Furthermore, we have theoretically proven that queries based on this index are accurate, and statistical analysis is feasible. Both algorithms have a time complexity of $O(N)$ in indexing construction, demonstrating good performance.

As societies evolve, wireless sensing technologies have rapidly matured and now serve as indispensable building blocks in fields ranging from ecological surveillance and intelligent urban administration to medical monitoring and supply-chain oversight. These networks demand the ubiquitous scattering of tiny nodes that, without human oversight, perpetually harvest, process and relay information toward base stations or sinks. Within such architectures, the choice of routing strategy is critical; achieving uniform and minimal power depletion across nodes during multi-hop transfer has therefore emerged as a central research theme. Reinforcement learning-an experiential decision-making framework-offers a promising route to this goal. Accordingly, this paper introduces a power-aware routing protocol that embeds RL principles at every node. Each sensor independently runs a Q-learning routine to score its one-hop peers; the score consolidates four local indicators: the Euclidean span and angular bearing of the prospective link, the neighbour’s residual battery, and the power it expended when last forwarding a packet. Simulation results reveal that the proposed scheme attains lower and more balanced power dissipation than existing distributed alternatives.

Traffic flow prediction is a prominent research area in intelligent transportation systems, significantly contributing to urban traffic management and control. Existing methods or models for traffic flow prediction predominantly rely on a fixed-graph structure to capture spatial correlations within a road network. However, the fixed-graph structure can restrict the representation of spatial information due to varying conditions such as time and road changes. Drawing inspiration from the attention mechanism, a new prediction model based on the mixed-graph neural network is proposed to dynamically capture the spatial traffic flow correlations. This model uses graph convolution and attention networks to adapt to complex and changeable traffic and other conditions by learning the static and dynamic spatial traffic flow characteristics, respectively. Then, their outputs are fused by the gating mechanism to learn the spatial traffic flow correlations. The Transformer encoder layer is subsequently employed to model the learned spatial characteristics and capture the temporal traffic flow correlations. Evaluated on five real traffic flow datasets, the proposed model outperforms the state-of-the-art models in prediction accuracy. Furthermore, ablation experiments demonstrate the strong performance of the proposed model in long-term traffic flow prediction.

In recent years, point-of-interest (POI) recommendation has been extensively studied, with existing methods typically modeling user preferences through the integration of multi-factor information (e.g., temporal, spatial, and categorical features) and capturing the periodicity and discontinuity of user check-in sequences. However, these approaches struggle with data sparsity, missing data, and noisy data, leading to suboptimal user representations. To effectively mitigate the above problems, we utilize contrastive self-supervised learning techniques to achieve data augmentation and apply them to the next POI recommendation task. Specifically, we propose DACL (Data Augmentation through Contrastive Self-supervised Learning), a novel framework that unifies next POI recommendation and contrastive self-supervised learning (SSL) via a multi-task strategy. Furthermore, DACL introduces five tailored data augmentation operations to generate high-quality contrastive views, mitigating data limitations while enhancing robustness. Extensive experiments on two real-world datasets (NYC and TKY) demonstrate that DACL significantly outperforms state-of-the-art baselines, achieving 14.3% and 4.3% improvements in Recall@10 and NDCG@5, respectively, while maintaining superior robustness against noisy and sparse scenarios.

Spatial distribution similarity analysis has extensive application value in multiple domains including geographic information science, urban planning, and engineering site selection. However, traditional regional similarity analysis methods face three key challenges: high sensitivity to directional changes, limitations in feature interpretability, and insufficient adaptability to multi-type data. Addressing these issues, this paper proposes a rotation-invariant spatial distribution similarity analysis method based on ring vectors. This method comprises three stages. First, the traversal starting point of the ring vector is dynamically selected based on the maximum value point of the regional feature matrix. Next, concentric ring features are extracted according to this starting point to achieve multi-scale characterization. Finally, the bidirectional weighted comprehensive distance of ring vectors between regions is calculated to measure the similarity between regions. Three experimental sets verified the method’s effectiveness in terrain matching, engineering site selection, and urban functional area identification. These results confirm its rotational invariance, feature interpretability, and adaptability to multi-type data. This research provides a new technical approach for spatial distribution similarity analysis, with significant theoretical and practical implications for geographic information science, urban planning, and engineering site selection.

With the increase in urban traffic congestion, the problem of identifying vulnerable areas of road networks has attracted wide attention. Existing studies have mainly used mediated centrality methods or algorithms based on link analysis to find critical and vulnerable nodes in road network systems through network centrality methods. However, the experimental results of the mediated centrality approach under large-scale road network datasets have a certain degree of randomness, while the link analysis algorithms ignore the structure and interrelationships among nodes. For this reason, this paper proposes a new method to quickly identify vulnerable areas of road networks based on interflow degree and spatial density clustering. The main contribution of this paper is the introduction of a clustering-based road network density-related area vulnerability indicator $AVI$ aimed at assessing the level of area vulnerability. The indicator focuses on the relationship between the road network structure and nodes, avoiding randomness by considering all intersection nodes. Theoretical analysis indicates that the proposed method demonstrates excellent global stability and achieves a significantly lower time complexity of $O(n^{2})$ compared to the traditional Betweenness Centrality method, which has a time complexity of $O(n^{5})$ . Through experimental validation of data from Beijing, London, New York, and Tokyo, the results show that the proposed method has high accuracy and reliability in identifying vulnerable areas of the road network with acceptable time complexity.

With the vigorous development of transportation infrastructure in various countries, the traffic network within the city is becoming more and more complex, and when an emergency occurs in one or more areas of the city, it will inevitably cause traffic congestion in the area and keep spreading. There are still many challenges to solve the urban emergency route planning problem. In this paper, we have employed a double layer search structure, where we have empowered the traditional A* model with a neural network, to construct a region-level dynamic path planning model known as “Double Layer A*”. The model divides the road network into two layers, and implements the outer layer and inner layer search. In the outer layer search, we use the historical cab travel data for training to achieve the general direction planning; in the inner layer search, we update the original planning according to the changes of the road condition characteristics of the regional nodes, and perform the re-planning in real time. We conducted experimental evaluations using the road network data of Beijing, and the results showed that compared to a single-layer search structure path planning model, our Double layer A* model planned paths with higher similarity in land characteristics, connectivity, and average connectivity between adjacent nodes, which demonstrates the effectiveness and reasonableness of the Double layer A* model in emergency path planning.

Modern processors employ data prefetchers to alleviate the impact of long memory access latency. However, current prefetchers are designed for specific memory access patterns, which perform poorly on mixed applications with multiple memory access patterns. To address these issues, RL-CoPref, a reinforcement learning (RL)-based coordinated prefetching controller for multiple prefetchers, is proposed in this paper. RL-CoPref takes diverse program context information as the input, learns to maximize cumulative rewards, and evaluates prefetch quality based on prefetch hits/misses and memory bandwidth utilization. It can dynamically adjust the prefetch activation and prefetch degree, enabling multiple prefetchers to complement each other on mixed applications. Our extensive evaluation, utilizing the ChampSim simulator, demonstrates that RL-CoPref can effectively adapt to various workloads and system configurations, optimizing prefetch control. On average, RL-CoPref achieves 76.15% prefetch coverage, having 35.50% IPC improvement, outperforming state-of-the-art individual prefetchers by 5.91–16.54% and outperforming SBP, a state-of-the-art (non-RL) prefetch controller, by 4.64%.

Modern data centers provide the foundational infrastructure of cloud computing. Workload generation, which involves simulating or constructing tasks and transactions to replicate the actual resource usage patterns of real-world systems or applications, plays essential role for efficient resource management in these centers. Data center traces, rich in information about workload execution and resource utilization, are thus ideal data for workload generation. Traditional traces provide detailed temporal resource usage data to enable fine-grained workload generation. However, modern data centers tend to favor tracing statistical metrics to reduce overhead. Therefore the accurate reconstruction of temporal resource consumption without detailed, temporized trace information become a major challenge for trace-based workload generation. To address this challenge, we propose STWGEN, a novel method that leverages statistical trace data for workload generation. STWGEN is specifically designed to generate the batch task workloads based on Alibaba trace. STWGEN contains two key components: a suite of C program-based flexible workload building blocks and a heuristic strategy to assemble building blocks for workload generation. Both components are carefully designed to reproduce synthetic batch tasks that closely replicate the observed resource usage patterns in a representative data center. Experimental results demonstrate that STWGEN outperforms state-of-the-art workload generation methods as it emulates workload-level and machine-level resource usage in much higher accuracy.

Polyurethane elastomers address important significance in a wide range of applications. The mechanical properties including strength and toughness are of great importance, which are closely related to their unique hydrogen bonding structure. Unfortunately, the poor designability of hydrogen bonding structure in existing polyurethane severely restricts the on-demand regulation of their properties. Herein, a facile, universal and efficient modifying strategy based on stimuli-responsive polyphenol aggregates was proposed. Through precisely manipulated heat-induced aggregate division and/or photo-induced interfacial hydrogen bonding upgrading, programmable strengthening and toughening effect on the derived polyurethane elastomers could be achieved with high precision. Typically, the tensile strength and toughness of our proposed polyurethane elastomers could be enhanced by 3.23 and 2.22 times comparing with neat samples, respectively. The relevant results were supported by various characterizations and mathematical modeling. In addition, our polyurethane exhibited unique selective biocompatibility, rapid self-healing capability and recyclability, which could fulfill varieties of functions. Our proposed modifying strategy by using polyphenol aggregates can not only programmatically optimize the comprehensive properties of polyurethane, but also inspire programmable regulation of polymer performance through programmable design of its certain microstructure in the future.

In the era of big data, unsupervised learning algorithms such as clustering are particularly prominent. In recent years, there have been significant advancements in clustering algorithm research. The Clustering by Density Peaks algorithm is known as Clustering by Fast Search and Find of Density Peaks (density peak clustering). This clustering algorithm, proposed in Science in 2014, automatically finds cluster centers. It is simple, efficient, does not require iterative computation, and is suitable for large-scale and high-dimensional data. However, DPC and most of its refinements have several drawbacks. The method primarily considers the overall structure of the data, often resulting in the oversight of many clusters. The choice of truncation distance affects the calculation of local density values, and varying dataset sizes may necessitate different computational methods, impacting the quality of clustering results. In addition, the initial assignment of labels can cause a ‘chain reaction’, i.e., if one data point is incorrectly labeled, it may lead to more subsequent data points being incorrectly labeled. In this paper, we propose an improved density peak clustering method, DPC-MS, which uses the mean-shift algorithm to find local density extremes, making the accuracy of the algorithm independent of the parameter dc. After finding the local density extreme points, the allocation strategy of the DPC algorithm is employed to assign the remaining points to appropriate local density extreme points, forming the final clusters. The robustness of this method in handling uncertain dataset sizes adds some application value, and several experiments were conducted on synthetic and real datasets to evaluate the performance of the proposed method. The results show that the proposed method outperforms some of the more recent methods in most cases.

Traffic congestion in urban areas has become a major worldwide problem. As an important direction of the Intelligent Transportation System (ITS), traffic-speed prediction can help drivers better plan routes and shorten travel time according to IOT techniques, thereby effectively alleviating the problem of traffic congestion. Traffic speed changes dynamically over time, so forecasting using historical data may not be able to quickly adapt to sudden changes in traffic conditions, and predicted traffic conditions may lag behind. The use of real-time data can capture instantaneous changes in traffic conditions, which is more adaptable to different traffic scenarios. In order to further explore the advantages of using real-time data for traffic forecasting, a novel real-time Data Driven method for traffic Speed trend Prediction (2DSP) is proposed. The 2DSP method can predict the macro-level traffic speed trend (rising or falling) by using only near-real-time microscopic vehicle information, which effectively captures the dynamic changes in traffic speed. In addition, an adaptive time-slicing strategy based on traffic density is proposed. This strategy dynamically divides time slices based on traffic density, reducing the frequency of data processing and improving the user experience. The effectiveness of the 2DSP method is validated using two real traffic datasets of floating vehicles. The experimental results show that the 2DSP method has good potential for real-time traffic-speed trend prediction.

The automatic price evaluation is an important and challenging issue for the used mobile phone recycling. However, due to complicated relationships between attributes and price as well as insufficient sample data, current approaches cannot achieve accurate price evaluation for the used mobile phones, which significantly reduce the efficiency for the used mobile phone recycling. To this end, a tensor based approach is proposed in this paper, which can achieve accurate and efficient price evaluation for used mobile phones. In the proposed approach, first, the mutual information based attributes selection mechanism and the boxplot based sample selection mechanism are employed to select the most relevant attributes and suitable price samples from used mobile phone dataset. Then, a tensor model is constructed to establish multi-dimensional relationships between selected attributes and prices of used mobile phones. Finally, the missing values in the constructed tensor model is completed through the gradient descent and CANDECOMP/PARAFAC decomposition algorithms. The completed tensor model can be directly used for the price evaluation of used mobile phones based on their attributes without further calculation. Based on the real price dataset of used mobile phones, the experiments indicate that the proposed approach outperforms most of current approaches for the price evaluation for the used mobile phones.

When multiple central processing unit (CPU) cores and integrated graphics processing units (GPUs) share off-chip main memory, CPU and GPU applications compete for the critical memory resource. This causes serious resource competition and has a negative impact on the overall performance of the system. We describe the competition for shared-memory resources in a CPU-GPU heterogeneous multi-core architecture, and a shared-memory request scheduling strategy based on perceptual and predictive batch-processing is proposed. By sensing the CPU and GPU memory request conditions in the request buffer, the proposed scheduling strategy estimates the GPU latency tolerance and reduces mutual interference between CPU and GPU by processing CPU or GPU memory requests in batches. According to the simulation results, the scheduling strategy improves CPU performance by 8.53% and reduces mutual interference by 10.38% with low hardware complexity.

Missing traffic data collected by IoT sensors is a common issue. Having complete traffic data can help people with their studies and work in real world. A spatio-temporal enhanced k nearest neighbor (ST-KNN) method is proposed in this paper to interpolate missing traffic data according to its corresponding spatio-temporal dependence. The proposed method is improved in three aspects: initially, localized data are involved in the computation, the distance metric formula is re-designed secondly, and the data regression model is improved. We conducted our experimental evaluations on publicly available real dataset, and the results are compared to those from existing state-of-the-art models. The results of our experiments show that the method proposed in this paper can effectively improve traffic data interpolation accuracy.

Urban rail transit is an essential part of the urban public transportation system. The reasonable spatial data visualization of urban rail transit stations can provide a more intuitive way for the majority of travelers to arrange travel plans and find destinations. The map service of rail transit stations generated by data visualization has gradually become indispensable information guidance in the rail transit system. The existing map service icons block each other when the scale changes, and new stations cannot be displayed dynamically when users drag the map. This paper uses filtering and sorting methods to dynamically query and visualize the relatively more important transportation stations within the users’ visible range, so as to solve the above problems and provide people with better transportation services. Our method introduces three constraints: spatial diversity, time-sharing passenger flow analysis and whether it is a transit station, and calculates the scores of constraint relationships of feature objects to evaluate stations. On the basis of the skyline query, the scores of feature objects are combined and sorted to obtain an ordered object set of the most interesting k points(top-k POIs), and the rail transit stations are dynamically retrieved and visualized. Before sorting POIs, we filter out POIs that need to be fitted, so that only the k most representative POIs in the currently visible range are displayed. When the map scale changes, the displayed POIs are updated. Finally, through the statistics of efficiency calculation of this method under different scales and centers, combined with users’ evaluations, it was proved that our method could better display critical information and improve user experience.