To ensure reliability, modern deduplication-based storage systems exploit erasure coding to redundantly distribute objects that store post-deduplicated data across multiple storage nodes. In the case of node failures, the failure recovery of impacted objects performs degraded read operations by reading from these storage nodes. However, this degraded read scheme introduces the additional available objects into the restore cache, which poses challenges to state-of-the-art caching policies applied in erasure-coded deduplicated storage systems. On one hand, chunk-based caching method fails to cache the remaining available objects after the unavailable objects are rebuilt, resulting in multiple reads of accessed objects and severely degrading restore performance. On the other hand, object-based caching methods do not identify lost objects to which duplicate chunks belong, causing unused objects to unnecessarily occupy cache space and destroy cache locality. In this paper, we propose TC-Cache, a restore cache scheme for node failures, which combines object-based caching and chunk-based caching methods to improve restore performance. The main idea of TC-Cache is to identify objects to which redundant data belongs, and selectively handle degraded reads of two types of objects in caches of different granularities during failure recovery to enhance cache utilization and maintain cache locality. Extensive experiments based on three real-world datasets demonstrate that TC-Cache outperforms state-of-the-art cache algorithms in terms of restore performance by 21%-65% with low computing overhead.

Multimodal 3D segmentation involves a significant number of 3D convolution operations, which requires substantial computational resources and high-performance computing devices in MRI multimodal brain tumor segmentation. The key challenge in multimodal 3D segmentation is how to minimize network computational load while maintaining high accuracy. To address the issue, a novel lightweight parameter aggregation network (SuperLightNet) is proposed to realize the efficient encoder and decoder for the high accurate and low computation. A random multiview drop encoder is designed to learn the spatial structure of multimodal images through a random multi-view approach for solving the high computational time complexity that has arisen in recent years with methods relying on transformers and Mamba. A learnable residual skip decoder is designed to incorporate learnable residual and group skip weights for addressing the reduced computational efficiency caused by the use of overly heavy convolution and deconvolution decoders. Experimental results demonstrate that the proposed method achieves a leading reduction in parameter count by 95.59%, the 96.78% improvement in computational efficiency, the 96.86% enhancement in memory access performance, and the average performance gain of 0.21% on the BraTS2019 and BraTS2021 datasets in comparison with the state-of-the-art methods. Code is available at https://github.com/WTU-MIS-Laboratory/SuperLightNet.

Smart wearable systems can monitor users’ physiological data in real time, detect anomalies promptly through risk recognition technologies, provide early warnings, and assist users in taking preventive measures. However, single modal information is difficult to accurately recognize the behavioral state, expression state, and environmental conditions. Furthermore, multimodal data are often affected by noise and interference, complicating the accurate identification of risky behaviors. To address these challenges, we propose a smart wearable system based on the dual-frequency enhancement network (DFENet): 1) the multimodal sensor system is designed to combine behavioral recognition, expression recognition, and environmental recognition for comprehensive monitoring and recognition of multidimensional risk factors in complex scenarios; 2) the DFENet is proposed to overcome challenges in feature extraction and accurate classification in complex environments; and 3) the behavioral recognition dataset and the expression recognition dataset are built to verify the effectiveness of the designed smart wearable system. Experimental results indicate that the proposed system can real-time achieve risk recognition across physical activity, expression state, and environmental conditions, and the proposed DFENet achieves excellent performance in accuracy, parameters, and floating-point operations (FLOPs) metrics on the three datasets. The algorithm and datasets can be downloaded athttps://github.com/wtu1020/Multimodal-Wearable.

Brain-computer interface (BCI) offers a direct communication and control channel between the human brain and external devices, presenting new pathways for individuals with physical disabilities to operate robotic arms for complex tasks. However, achieving multiobject grasping tasks under low signal-to-noise ratio (SNR) consumer-grade EEG signals is a significant challenge due to the lack of robust decoding algorithms and precise visual tracking methods. This article proposes, ArmBCIsys, an integrated robotic arm system that combines a novel dual-branch frequency-enhanced network (DBFENet) to robustly decode EEG signals under noisy conditions with the high-precision vision-guided grasping module. The proposed DBFENet designs the scaling temporal convolution block (STCB) to extract multiscale spatiotemporal features from the time domain, while the designed DropScale projected Transformer (DSPT) utilizes discrete cosine transform (DCT) to capture key frequency-domain features, significantly improving decoding robustness. We fine-tune the masked-attention mask Transformer (Mask2Former) model on the Jacquard dataset and incorporate the multiframe centroid-intersection over union (IoU) tracking algorithm to build visual grasp segmenter (VisGraspSeg), enabling reliable segmentation and dynamic tracking for diverse daily objects. Experimental validations on both self-built code-modulated visual evoked potential (c-VEP) dataset (1344 samples) and two public c-VEP datasets demonstrate that DBFENet achieves the state-of-the-art recognition performance, and the system integrates the DBFENet and proposed vision-guided module and ensures stable multiobject selecting and automatic object grasping in dynamic environments, extending promising applications in healthcare robotics, assistive technology, and industrial automation. The self-built dataset has been made publicly accessible at https://github.com/wtu1020/ ArmBCIsys-Self-built-cVEP-Dataset.

Intelligent wearable systems have been widely used in health monitoring, motion tracking, and engineering safety. However, the single function of current wearable systems cannot satisfy the requirements of complex scenarios, and the wearable systems cannot establish a relationship with the virtual 3D visualization platform. To address these issues, this paper proposes a novel intelligent wearable system with motion and emotion recognition. Multiple sensors are integrated into the system to collect motion and emotion information. In order to achieve accurate classification and recognition of multiple sensor information, we propose a novel human action recognition network called the three-branch spatial-temporal feature extraction network (TB-SFENet), which can obtain more robust features and achieve an accuracy of 97.04% on the UCI-HAR dataset and 92.68% on the UniMiB SHAR dataset. To establish the relationship between the real entity and virtual space, we use digital twin (DT) technology to establish the 3D display DT platform. The platform enables real-time information interaction, such as activity, emotion, location, and monitoring information. Additionally, we establish the TGAM electroencephalogram emotion classification (TEEC) dataset, which contains 120,000 pieces of data, for the proposed system. Experimental results indicate that the proposed system realizes virtual reality information interaction between the personal digital human and actual person based on the intelligent wearable system, which has great potential for applications in intelligent healthcare, virtual reality, and other fields.

Abstract Wearable human action recognition (HAR) has practical applications in daily life. However, traditional HAR methods solely focus on identifying user movements, lacking interactivity and user engagement. This paper proposes a novel immersive HAR method called MovPosVR. Virtual reality (VR) technology is employed to create realistic scenes and enhance the user experience. To improve the accuracy of user action recognition in immersive HAR, a multi‐scale spatio‐temporal attention network (MSSTANet) is proposed. The network combines the convolutional residual squeeze and excitation (CRSE) module with the multi‐branch convolution and long short‐term memory (MCLSTM) module to extract spatio‐temporal features and automatically select relevant features from action signals. Additionally, a multi‐head attention with shared linear mechanism (MHASLM) module is designed to facilitate information interaction, further enhancing feature extraction and improving accuracy. The MSSTANet network achieves superior performance, with accuracy rates of 99.33% and 98.83% on the publicly available WISDM and PAMPA2 datasets, respectively, surpassing state‐of‐the‐art networks. Our method showcases the potential to display user actions and position information in a virtual world, enriching user experiences and interactions across diverse application scenarios.

Abstract Human activity recognition (HAR) has significant potential in virtual sports applications. However, current HAR networks often prioritize high accuracy at the expense of practical application requirements, resulting in networks with large parameter counts and computational complexity. This can pose challenges for real‐time and efficient recognition. This paper proposes a hybrid lightweight DSANet network designed to address the challenges of real‐time performance and algorithmic complexity. The network utilizes a multi‐scale depthwise separable convolutional (Multi‐scale DWCNN) module to extract spatial information and a multi‐layer Gated Recurrent Unit (Multi‐layer GRU) module for temporal feature extraction. It also incorporates an improved channel‐space attention module called RCSFA to enhance feature extraction capability. By leveraging channel, spatial, and temporal information, the network achieves a low number of parameters with high accuracy. Experimental evaluations on UCIHAR, WISDM, and PAMAP2 datasets demonstrate that the network not only reduces parameter counts but also achieves accuracy rates of 97.55%, 98.99%, and 98.67%, respectively, compared to state‐of‐the‐art networks. This research provides valuable insights for the virtual sports field and presents a novel network for real‐time activity recognition deployment in embedded devices.

Intelligent garments, a burgeoning class of wearable devices, have extensive applications in domains such as sports training and medical rehabilitation. Nonetheless, existing research in the smart wearables domain predominantly emphasizes sensor functionality and quantity, often skipping crucial aspects related to user experience and interaction. To address this gap, this study introduces a novel real-time 3D interactive system based on intelligent garments. The system utilizes lightweight sensor modules to collect human motion data and introduces a dual-stream fusion network based on pulsed neural units to classify and recognize human movements, thereby achieving real-time interaction between users and sensors. Additionally, the system in- corporates 3D human visualization functionality, which visualizes sensor data and recognizes human actions as 3D models in realtime, providing accurate and comprehensive visual feedback to help users better understand and analyze the details and features of human motion. This system has significant potential for applications in motion detection, medical monitoring, virtual reality, and other fields. The accurate classification of human actions con- tributes to the development of personalized training plans and injury prevention strategies. This study has substantial implications in the domains of intelligent garments, human motion monitoring, and digital twin visualization. The advancement of this system is expected to propel the progress of wearable technology and foster a deeper comprehension of human motion.

Anomaly detection system plays an important role in maintaining the stability of cloud database. Existing studies mainly focus on significant deviations in multivariate time series, such as a combination of CPU utilization, transactions per second, etc, to detect abnormal issues. Due to the complexity of cloud database structure and functions, these approaches are difficult to achieve a balance among detection performance, detection efficiency and workload adaptability. In this paper, we propose DBCatcher, a cloud database online anomaly detection system based on indicator correlation. Through extensive analysis of real-world cloud database time series, we find the correlations among trends in the same key performance indicators across databases within the same unit, which inspires us to explore a time series correlation measurement method that can efficiently detect abnormal issues. Meanwhile, we design a flexible time window observation mechanism and an adaptive threshold learning policy to minimize misjudgment caused by key performance indicator fluctuations, greatly enhancing the detection performance and workload adaptability. We conduct extensive experiments under real-world and synthetic workloads. Experimental results show that DBCatcher significantly improves the detection performance and detection efficiency compared to existing methods.

In the fashion field, with the increasing diversity of clothing types and styles, accurate clothing classification becomes very important. However, the complex background and diverse styles of clothing images bring challenges to feature extraction. Classification based on texture features alone may focus too much on details and ignore the overall shape information, thus reducing the accuracy and stability of classification. In order to achieve fast and accurate clothing classification, this paper proposes a two-stream network structure clothing classification algorithm based on shape texture features and multi-feature fusion (TSFCloNet). Its main core is as follows: 1) using the two-stream network structure to extract texture and shape features from the input data set respectively; 2) in the shape feature extraction stream, the clothing shape acquisition module is first used to process the input clothing data set, and the obtained clothing shape data set is input into the ShapeNet feature extraction module to obtain shape feature information; 3) the FFCE (Feature Fusion Channel Enhancement) module is used to fuse the features obtained by the two branches of the structure respectively, and the DSAConv module is used to enhance feature extraction, and the final features are sent to the trained classifier to obtain the clothing style classification results. A large number of experimental results show that the proposed TSFCloNet network achieves higher classification accuracy when dealing with diverse and changeable fashion styles, significantly improving the performance of fashion image classification.

File storage system (FSS) uses multi-caches to accelerate data accesses. Unfortunately, efficient FSS cache allocation remains extremely difficult. First, as the key of cache allocation, existing miss ratio curve (MRC) constructions are limited to LRU. Second, existing techniques are suitable for same-layer caches but not for hierarchical ones.

Index structure is very important for efficient data access and system performance in the storage system. Learned index utilizes recursive index models to replace range index structure (such as B+ Tree) so as to predict the position of a lookup key in a dataset. This new paradigm greatly reduces query time and index size, however it only supports read-only workloads. Although some studies reserve gaps between keys for new data to support update, they incur high memory space and shift cost when a large number of data are inserted.

Recently, using machine learning for performance tuning of cloud database (CDB) service has shown great potentials. However, facing personalized requirements such as various restrictions for tuning with very different workloads, pre-trained models may mismatch or recommend suboptimal configurations given a new workload. On the other hand, if the system tunes configurations in an online fashion, the system will suffer from the cold start problem, resulting in long tuning time and performance fluctuation. To accommodate these problems, we propose an online CDB tuning system called HUNTER. The key feature of HUNTER is a hybrid architecture, which uses samples generated by Genetic Algorithm to warm-start the finer grained exploration of deep reinforcement learning. Meanwhile, we employ Principal Component Analysis, Random Forest, and Fast Exploration Strategy to reduce the search space and the update time of the learning model. In addition, we further propose a clone and parallelization scheme to stress-test workloads on multiple cloned CDB instances (CDBs), resulting in faster and safer configuration exploration. Extensive trials on CDB with public and real-world workloads demonstrate that, given the same time budget and resources, HUNTER improves performance and considerably decreases recommendation time compared to state-of-the-art tuning systems, with accelerations of up to 2.8× and 22.8× utilizing 1 and 20 cloned CDBs, respectively.

The popular Log-Structured Merge (LSM) tree based Key-Value (KV) stores make trade-offs between write cost and read cost via different merge policies, i.e., leveling and tiering. It has been widely documented that leveling severely hampers write throughput, while tiering hampers read throughput. The characteristics of modern workloads are seriously challenging LSM-tree based KV stores for high performance and high scalability on SSDs. In this work, we present PTierDB, an LSM-tree based KV store that strikes the better balance between read cost and write cost for small data on SSD via an adaptive tiering principle and three merge policies in the LSM-tree, leveraging both the sequential and random performance characteristics of SSDs. Adaptive tiering introduces two merge principles: prefix-based data split which bounds the lookup cost and coexisted merge and move which reduces data merging. Based on adaptive tiering, three merge policies make decisions to merge-sort or move data during the merging processes for different levels. We demonstrate the advantages of PTierDB with both microbenchmarks and YCSB workloads. Experimental results show that, compared with state-of-the-art KV stores and the KV implementations with popular merge policies, PTierDB achieves a better balance between read cost and write cost, and yields up to 2.5x improvement in the performance and 50% reduction of write amplification.

Analyzing large-scale image dataset requires all images to be read from disks first, leading to high read latency. Therefore, we propose a Content Sifting Storage (CSS) system, which aims to reduce the read latency by only reading sifted relevant data. CSS generates embedded content metadata via deep learning and manages the metadata via Semantic Hamming Graph, which achieves fast read based on content similarity meeting the given analysis. Extensive experimental results on image datasets show that compared with conventional semantic storage systems, our CSS can greatly reduce the read latency by 82.21% to 94.8% with more than 98% recall rate.

Configuration tuning is vital to optimize the performance of database management system (DBMS). It becomes more tedious and urgent for cloud databases (CDB) due to the diverse database instances and query workloads, which make the database administrator (DBA) incompetent. Although there are some studies on automatic DBMS configuration tuning, they have several limitations. Firstly, they adopt a pipelined learning model but cannot optimize the overall performance in an end-to-end manner. Secondly, they rely on large-scale high-quality training samples which are hard to obtain. Thirdly, there are a large number of knobs that are in continuous space and have unseen dependencies, and they cannot recommend reasonable configurations in such high-dimensional continuous space. Lastly, in cloud environment, they can hardly cope with the changes of hardware configurations and workloads, and have poor adaptability. To address these challenges, we design an end-to-end automatic CDB tuning system, CDBTune, using deep reinforcement learning (RL). CDBTune utilizes the deep deterministic policy gradient method to find the optimal configurations in high-dimensional continuous space. CDBTune adopts a try-and-error strategy to learn knob settings with a limited number of samples to accomplish the initial training, which alleviates the difficulty of collecting massive high-quality samples. CDBTune adopts the reward-feedback mechanism in RL instead of traditional regression, which enables end-to-end learning and accelerates the convergence speed of our model and improves efficiency of online tuning. We conducted extensive experiments under 6 different workloads on real cloud databases to demonstrate the superiority of CDBTune. Experimental results showed that CDBTune had a good adaptability and significantly outperformed the state-of-the-art tuning tools and DBA experts.

Log-Structured Merge Trees (LSM) based key-value (KV) stores such as LevelDB and HyperLevelDB, use a compaction strategy which brings frequent compaction operations, to store key-value items in sorted order. However, large numbers of compactions impose a negative impact on write and read performance for random data-intensive workloads. To remedy this problem, this paper presents OHDB, an optimization of HyperLevelDB for random data-intensive workloads. OHDB implements two stand-alone techniques in the disk component of LSM structure to optimize the concurrent compactions. One is dividing KV items by prefix at the first level in the disk component, to reduce the frequency of overlapping in key range among data files, and thus reduces the amount of compactions. The other is separating the first level in the disk component from the rest levels, and organizing them in two disks individually, to increase parallelism of disk writes of compactions. We evaluate three OHDBs which are OHDB with each of the technique and OHDB with the combination of both respectively, using micro-benchmarks with random write- intensive and read-intensive workloads. Experimental results show that OHDB reduces the amount of compactions by a factor of up to 4x, and improves the write and read performance for random data-intensive workloads under various settings.