ArchiveWith rapid advancements in edge computing, sensing, AI, and communication technologies, vehicles are undergoing an unprecedented transformation. We introduce a new computing paradigm for the autonomous driving era—vehicle computing. In this paradigm, data and control layers are separated, creating an open computing platform that supports multi-party collaboration and data sharing, breaking away from the limitations of traditional, closed vehicle systems. This paradigm enables vehicles to transcend conventional transportation roles, evolving into versatile mobile computing platforms that support a wide range of advanced applications and third-party services. We define the core concepts of vehicle computing, analyze the revolutionary evolution of software and computing architectures within vehicles, and present promising application examples, as well as a novel business model enabled by this paradigm. The five core functionalities of vehicle computing, i.e., computation, communication, energy management, sensing, and data storage, and their related cutting-edge technologies are thoroughly explored. We conclude by discussing key technical challenges and promising opportunities within vehicle computing, aiming to inspire further academic and industry research in this innovative field.
Central processing unit is the most important computing infrastructure nowadays. To maximize the profit, architects design the processor microarchitecture by trading-off multiple objectives including performance, power, and area. However, because of the tremendous instructions of workloads running on the processors, the evaluation of individual microarchitecture design point costs minutes to hours. Furthermore, the design space of the microarchitecture is huge, which results that the exploration of comprehensive design space is unrealistic. Therefore, many machine-learning-assisted design space exploration acceleration methods are proposed to reduce the size of evaluated design space or accelerate the evaluation of a design point. However, a comprehensive survey summarizing and systematically classifying recent acceleration methods is missing. This survey paper systematically summarizes and classifies the five kinds of acceleration methods for the design space exploration of the processor microarchitecture, including the workload selection of software design space, the partial simulation of workload instructions, the design point selection, the simulation tools, and the performance models. This paper systematically compares the similarities and differences between papers in the acceleration methods, and covers the complete exploration process from the software workload selection to the hardware microarchitecture design. Finally, the research direction is summarized, and the future development trend is discussed.
NAND flash is widely utilized in mobile devices due to its excellent characteristics, including large capacity, light weight, and shock resistance. The flash friendly file system (F2FS), designed for flash features, is a typical log-structured file system (LFS). It employs a log-structured write mechanism to enhance random write performance, utilizes roll-forward recovery technology for fast consistency protection, and is commonly used as a file system for mobile devices. However, the performance of file system is impacted by fragmentation and segment cleaning. The out-of-place update mechanism of LFS and the small write mode of high-concurrency and random synchronization of mobile applications exacerbate fragmentation, leading to sluggish I/O request responses and device operation freezes. We initially introduce the relevant concepts and content of log-structured file systems in mobile devices. We then primarily outline the research status of fragmentation and segment cleaning of LFS. Firstly, we analyze the generation and impact of fragmentation, summarize the research work on reducing fragments from the perspectives of preventing fragments and reorganizing fragments. Secondly, we examine the impact of the mixed storage of hot and cold data on segment cleaning. Additionally, we summarize the research status of distinguishing hot and cold data from static and dynamic classification, and segment cleaning from the perspectives of managing data distribution and adjusting the timing, frequency, and objects of segment cleaning. Finally, we outline the main challenges and future research prospects of log-structured file systems in mobile devices.
While vector processing unit is widely employed in processors for neural networks, signal processing, and high performance computing, it suffers from expensive shuffle operations dedicated to data alignment. Traditionally, processors handle shuffle operations with its data shuffle unit. However, data shuffle unit will introduce expensive overhead of data movement and only can shuffle data in serial. In fact, shuffle operations only change the layout of data and ideally should be done entirely within memory. Nowadays, SRAM is no longer just a storage component, but also as a computing unit. To this end, we propose Shuffle-SRAM in this paper, and Shuffle-SRAM can shuffle multiple data elements simultaneously bit by bit within an SRAM bank. The key idea is to exploit the bit-line wise data movement ability in SRAM so as to shuffle multiple data in parallel, where all the bits of different data elements on the same bit-line of SRAM can be shuffled simultaneously, achieving a high level of parallelism. Through suitable data layout preparation and the vector shuffle extension instructions, Shuffle-SRAM efficiently supports a wide range of commonly used shuffle operations efficiently. Our evaluation results show that Shuffle-SRAM can achieve a performance gain of 28 times for commonly used shuffle operations and 3.18 times for real world applications including FFT, AlexNet, and VggNet. The SRAM area overhead only increases by 4.4%.
The graph data structure, which is adept at encapsulating intricate relationships among entities, has been widely used in a vast array of application scenarios. With the incessant progression of Internet applications and the concomitant surge in data scales, distributed graph computing systems have demonstrated superior performance compared with traditional single-machine systems in various aspects, including computational efficiency and resource scheduling. In recent years, the increasing demand for distributed graph computing systems designed for handling large-scale graph data has brought graph partitioning technology to the forefront of academic research. Based on a comprehensive analysis of graph partitioning techniques for distributed graph computing, we explain the technological backdrop of graph partitioning in these systems. We provide definitions for key concepts related to graph partitioning in modern distributed graph computing systems and present a classification scheme for existing computational models, offering insights into the current status of distributed graph computing paradigms. Subsequent sections delve into the complexities of different graph partitioning methodologies, conducting a thorough analysis to determine their respective strengths and weaknesses within the context of various distributed graph computing frameworks. Finally, we discuss the current challenges and future research directions of graph partitioning technology in distributed graph computing systems.
The feedback set problem is one of the most widely and deeply studied NP-complete graph problems in the field of computer science, with important applications in concurrent computing, large-scale integrated circuits, coding design, software verification, and social network analysis. The subset feedback set problem is a nature generalization of the feedback set problem, and has much universal and practical. In recent several years, the classification of the computational complexity for these two problems has drawn certain interests, and many breakthroughs have been made in the area of algorithms. In this paper, the survey on these problems mainly contains two parts. The first part introduces different versions of the feedback set and subset feedback set problems. The essential relations among these versions and their computational complexity in general graphs are also discussed in this part. The second part introduces the computational complexity of the feedback set and subset feedback set problems in some important and classical graph subclasses, including degree bounded graphs, planar graphs, tournaments, intersection graphs, forbidden graphs, and bipartite graphs. Finally, by analyzing and summarizing the existing research, the major research trends on the feedback set and subset feedback set problems are outlined.
Sequence alignment is a classical problem in biological sequence analysis, aiming to find out the similarity between sequences, which is of great significance for discovering functional, structural and evolutionary information in biological sequences. The problem can be divided into two categories: pairwise sequence alignment and multiple sequence alignment. The existing work is focused on specific algorithms, and no general solution is designed. In addition, there are few researches on the trustworthy of algorithms. By deeply analyzing the properties of sequence alignment problem and describing the essential characteristics of problem solving, a unified construction framework of sequence alignment dynamic programming algorithm seqAlign is designed based on the problem formal specification and formal method PAR. The process of constructing a multiple sequence alignment algorithm with three sequences by using the framework is further demonstrated, and the constructed results are formally verified by Isabelle theorem prover. Finally, the C++ executable program of the algorithm is generated by the code generation system of PAR platform. The process of mechanized construction of other sequence alignment algorithms using seqAlign framework is analyzed. Through strict specification refinement and formal verification, the reliability of the generated algorithm is effectively guaranteed. The developed seqAlign framework provides a general solution for the class of sequence alignment problems, which significantly improves the efficiency of generating sequence alignment algorithm families. The successful application of the designed seqAlign framework to sequence alignment problem in biological sequence analysis can provide a reference for the construction of highly reliable algorithms in complex bioinformatics field from the perspective of methodology and practice.
Point cloud segmentation algorithm based on deep learning can effectively segment point clouds in high-dimensional space by designing complex feature extraction modules. However, the lack of feature mining for boundary point set results in suboptimal accuracy in boundary segmentation. Some studies have applied the idea of contrastive learning to point cloud segmentation to solve the problem of insufficient boundary region segmentation performance, but the disorder and sparse characteristics of point cloud have not been fully utilized, and the feature extraction is not accurate enough. To solve these problems, we propose CL2M to learn more accurate features of point clouds at different locations through the self-attention mechanism, and the contrastive learning method is introduced to improve the segmentation accuracy of point cloud boundaries. In the process of contrastive boundary learning, labels in semantic space are deeply mined and a contrastive boundary learning module based on label distribution is designed to make the label distribution of point cloud in high-dimensional space contain more semantic information. The model makes full use of the label distribution law to calculate the distance between distributions, and can accurately divide positive and negative samples, reducing the cumulative errors caused by conventional hard partition. The results on two public data sets show that CL2M is superior to the existing point cloud segmentation model on several evaluation indexes, which verifies the effectiveness of the model.
Anomaly detection aims to detect abnormal samples among many normal samples. In the era of big data, how to apply anomaly detection to real-world scenarios has become one of the most critical problems to consider. Currently, the existing models can hardly cope with dynamic interference such as occlusion, lighting, and color difference in real-world scenarios and cannot quickly migrate application scenarios. We propose a deep learning model based on background-foreground modeling for anomaly detection tasks. Our model first reconstructs the input image into a clean background image without abnormal objects through the feature extraction network and preserves the possible dynamic interference of the image through skip-connection. After obtaining the reconstructed background, this model extracts the position information of abnormal objects through the spatial transformation network, uses an autoencoder to extract latent space representations of the appearance, shape, and presence information of abnormal objects, and reconstructs them. Finally, this model combines the reconstructed abnormal objects and the background image to obtain an overall reconstructed image and realizes anomaly detection by setting a threshold for the presence information of abnormal objects. To validate the effectiveness of the method, we collect data from a real circuit board assembly environment and simulate a scenario with limited annotations in actual production, resulting in the creation of a foreign object in circuit board (FO-CB) dataset for analysis. Additionally, we also conduct experimental validation on the foreign object debris in airport (FOD-A) dataset. The experimental results show that our proposed method performs well on the synthetic dataset and detects all anomalous objects in 9 actual scene data, with a miss rate of down to 0%, and can be applied to real-world circuit board assembly scenarios.
Graph contrastive learning is widely employed in recommender system due to its effectiveness in mitigating data sparsity issue. However, most current recommendation algorithms based on graph contrastive learning start to learn from only a single perspective, severely limiting the model’s generalization capability. Furthermore, the over-smoothing problem inherent in graph convolutional networks also affects the model’s stability. Based on this, we propose the multi-perspective graph contrastive learning recommendation method with layer attention mechanism. On the one hand, this method proposes three contrastive learning approaches from two different perspectives. From a view-level perspective, it constructs perturbation-enhanced view by adding random noise for the original graph and employing singular value decomposition (SVD) recombination to establish SVD-enhanced view. It then performs view-level contrastive learning on these two enhanced views. From a node-level perspective, it conducts contrastive learning on candidate nodes and candidate structural neighbors using semantic information between nodes, optimizes multi-task learning with three contrastive auxiliary tasks and a recommendation task to enhance the quality of node embeddings, thereby improving the model’s generalization ability. On the other hand, in the context of learning for user and item node embeddings by graph convolutional network, a layer attention mechanism is employed to aggregate the final node embeddings. This enhances the model’s higher-order connectivity and mitigates the over-smoothing issue. When compared with ten classic models on four publicly available datasets, such as LastFM, Gowalla, Ifashion, and Yelp, the results indicate that this method achieves an average improvement of 3.12% in Recall, 3.22% in Precision, and 4.06% in NDCG. This demonstrates the effectiveness of the approach proposed in this work.
The fusion of deep learning and the Internet of things has significantly promoted the development of the AIoT ecosystem. On the one hand, the huge amounts of multi-modal data collected by AIoT devices provide deep learning with abundant training data resources, which plays a more important role in the era of big models. On the other hand, the development of deep learning makes AIoT devices smarter, which shows great potential for promoting social development and the convenience of human life. As major support for the usage of deep learning in AIoT, federated learning effectively makes use of the training data provided by AIoT devices to train deep learning models with data privacy protection while collaborative inference overcomes the obstacles in the deployment of deep learning brought by the limited computation resource of AIoT devices. We induce the concept of AIoT-oriented collaborative intelligence. Aiming at implementing knowledge transmission and computation resource supply with high efficiency and security, we review the related works, published in the past 10 years, about the architecture, algorithm, privacy, and security of federated learning and collaborative inference, and introduce the inner connection of federated learning and collaborative inference. The algorithm part summarizes the federated learning and collaborative inference algorithm related to AIoT use cases and their optimization goals. The architecture part introduces the related works about deep learning accelerators, deep learning compilation, deep learning frameworks, communication among devices, and collaboration among devices from the view of AI computing systems. The privacy and security part introduces the privacy and security threats faced by AIoT-oriented collaborative intelligence and the defense methods against them. We also provide insights into the future development of AIoT-oriented collaborative intelligence in the aspect of equipment sharing, model sharing, collaboration of privacy and security mechanisms, and collaboration of incentive mechanisms.
Key performance indicator (KPI) anomaly detection is a fundamental technology for artificial intelligence for IT operations (AIOps) of Internet-based services. To improve the efficiency and accuracy of KPI anomaly detection, machine learning-based KPI anomaly detection has become a hotspot in both academia and industry recently. Through synthetically analyzing prior arts in this field, we first provide a technical framework of KPI anomaly detection for Internet-based services. Then, from the perspective of mining KPI’s dependency patterns in different domains (including time domain, metric domain and entity domain), we explore the motivation for model selection of KPI anomaly detection on three KPI types (including univariate KPI, multivariate KPIs and matrix-variate KPIs). Furthermore, guided by the detection performance objectives, we elaborate on KPI anomaly detection techniques from two perspectives: accuracy-centric anomaly detection techniques which focus on how to improve the accuracy of KPI anomaly detection models and multi-objective balancing-centric anomaly detection techniques which focus on how to balance theoretical performance with actual application objectives. Finally, we sort out five challenges on machine learning-based KPI anomaly detection, including KPI monitoring and KPI pre-processing, generality of the model, interpretability of the model, alarm management of anomalies, and limitations of KPI anomaly detection; and we also point out the corresponding potential research directions.
Wireless charging has the characteristics of no wiring, automation, and ease of maintenance, which can effectively solve the energy supply problem for the future large-scale application of artificial intelligence of things (AIoT). Magnetic resonant coupling wireless charging technology has more balanced technical advantages in terms of spatial freedom and energy transfer efficiency, and has become a research hotspot in the field of wireless charging today, with broader application prospects. In this paper, the development history of magnetic resonant coupling wireless charging technology is reviewed, and the related background concepts are briefly summarized. A systematic review of recent work on magnetic resonant coupling wireless charging technology is carried out, and based on the design motivation and technical realization, the basic principles and implementation methods of the corresponding research are summarized in detail from the four perspectives of feedback communication, optimization scheduling, sensing application and radiation safety guarantee, providing an effective reference for subsequent research. Specifically, feedback communication work is categorized into two main types according to transmission mode, optimization scheduling work is classified into two main types based on system structure, and sensing applications are introduced from three main categories according to application scenario. Finally, based on the aforementioned work, we discuss areas for improvement in the current magnetic resonant coupling wireless charging technology and explore future research directions.
Graph clustering based on local differential privacy has become a hot research topic in recent years. Most existing solutions are mainly realized through modular aggregation using bit vector technology. The linear relationship between the amount of noise and the vector dimension makes balancing clustering quality and privacy challenges. Aiming at the above problems, a directed graph clustering algorithm, DGC-LDP (directed graph clustering under LDP), is proposed based on edge local differential privacy (Edge-LDP). Concretely, the direct encoding method replaces the bit vector encoding method to reduce the amount of data in privacy processing. Meanwhile, a dynamic perturbation mechanism is designed based on the graph structure to balance privacy and statistical utility by adaptively adding noise. Then, according to the individual information uploaded by the terminal, the collector extracts the similarity information between nodes and designs a node aggregation algorithm based on the silhouette coefficient measurement model to generate clusters. Finally, an iterative mechanism is built between the terminal and the collector, and the collector iteratively optimizes the node aggregation form based on the statistical information fed back by the mechanism to achieve high-quality clustering. Theoretical analysis and experimentation on real-world datasets demonstrate that our proposed algorithm can obtain desirable clustering results while satisfying Edge-LDP.
In the era of big data e-commerce, data trading can enable collaborative sharing and value utilization of isolated data resources. As the main form of data trading in e-commerce business, marketing tags have enormous potential value. However, the traditional data trading market faces three main problems: 1) The opaque information of the centralized platform leads to trust crisis and malicious bidding ranking; 2) Lack of reasonable incentive mechanism to break the data island leads to data non-circulation and sharing difficulties; 3) Data security threats lead to privacy disclosure and data reselling and theft. In order to solve these problems, a blockchain marketing label trading mechanism for e-commerce alliance is designed, and the upper consensus incentive mechanism is designed based on decentralization, and all data transactions and computing businesses of the system are completed in combination with the trusted execution environment, thus realizing a safe and complete data transaction ecosystem. The authenticity verification mechanism is designed to ensure the effectiveness of marketing labels, the consensus incentive mechanism is designed to enable users to actively share data, and the smart contract is used to effectively constrain the role behavior according to the system design specification; Then key transmission and data security storage are realized through SGX remote authentication, and smart contract security call is realized to ensure user privacy and data security; Finally, the reliable delivery of data transaction results is realized through the trusted computer system and system design idea. In order to verify the security and practicability of the system, 350000 pieces of real data provided by an e-commerce company are used for performance testing. The test results show that the system can guarantee the security and performance requirements at the same time, and its additional costs mainly come from the remote authentication module and are within the acceptable range.