Journal of Systems Architecture最新文献_第7页

VM-PHRs: Efficient and verifiable multi-delegated PHRs search scheme for cloud–edge collaborative services vm - phr：针对云边缘协同服务的高效且可验证的多委托phr搜索方案

IF 4.1 2区计算机科学 Q1 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Journal of Systems Architecture

Pub Date : 2026-03-01 Epub Date: 2026-01-05 DOI: 10.1016/j.sysarc.2026.103689

Shiwen Zhang , Wenrui Zhu , Wei Liang , Arthur Sandor Voundi Koe , Neal N. Xiong

With the proliferation of smart healthcare services, many hospitals delegate PHRs processing to cloud-based resources. Despite its effectiveness for bounded search and selective record sharing over encrypted data, key-aggregate searchable encryption still suffers from significant drawbacks in current constructions. First, the existing trapdoor matching algorithms fail to achieve accurate matching and exhibit poor robustness against guessing attacks. Second, current works lack efficient mechanisms to enable fine-grained verification of search results. Third, there is currently no efficient mechanism to delegate user privileges. In this paper, we design an efficient and verifiable multi-delegated PHRs search scheme for cloud–edge collaborative services (VM-PHRs). To enable exact trapdoor matching and resist guessing attacks, we develop a new algorithm, EDAsearch. To achieve fine-grained verification of data integrity and correctness, we design a novel distributed protocol that operates over a network of edge servers. To accommodate real-world emergency scenarios, we develop a novel threshold mechanism that supports privilege delegation based on user attributes and hash commitments. Extensive security analysis and performance evaluation of VM-PHRs demonstrate that it is scalable, secure, and practical.

随着智能医疗保健服务的普及，许多医院将phrr处理委托给基于云的资源。尽管它在有界搜索和加密数据的选择性记录共享方面是有效的，但在当前的结构中，键聚合可搜索加密仍然存在明显的缺点。首先，现有的陷门匹配算法无法实现精确匹配，并且对猜测攻击的鲁棒性较差。其次，目前的工作缺乏有效的机制来实现对搜索结果的细粒度验证。第三，目前没有有效的机制来委派用户权限。本文针对云边缘协同服务（VM-PHRs），设计了一种高效、可验证的多委托PHRs搜索方案。为了实现精确的活板门匹配和抵抗猜测攻击，我们开发了一种新的算法，EDAsearch。为了实现对数据完整性和正确性的细粒度验证，我们设计了一个在边缘服务器网络上运行的新型分布式协议。为了适应现实世界的紧急情况，我们开发了一种新的阈值机制，该机制支持基于用户属性和哈希承诺的特权委托。VM-PHRs的广泛安全分析和性能评估表明，它具有可扩展性、安全性和实用性。

{"title":"VM-PHRs: Efficient and verifiable multi-delegated PHRs search scheme for cloud–edge collaborative services","authors":"Shiwen Zhang , Wenrui Zhu , Wei Liang , Arthur Sandor Voundi Koe , Neal N. Xiong","doi":"10.1016/j.sysarc.2026.103689","DOIUrl":"10.1016/j.sysarc.2026.103689","url":null,"abstract":"<div><div>With the proliferation of smart healthcare services, many hospitals delegate PHRs processing to cloud-based resources. Despite its effectiveness for bounded search and selective record sharing over encrypted data, key-aggregate searchable encryption still suffers from significant drawbacks in current constructions. First, the existing trapdoor matching algorithms fail to achieve accurate matching and exhibit poor robustness against guessing attacks. Second, current works lack efficient mechanisms to enable fine-grained verification of search results. Third, there is currently no efficient mechanism to delegate user privileges. In this paper, we design an efficient and verifiable multi-delegated PHRs search scheme for cloud–edge collaborative services (VM-PHRs). To enable exact trapdoor matching and resist guessing attacks, we develop a new algorithm, EDAsearch. To achieve fine-grained verification of data integrity and correctness, we design a novel distributed protocol that operates over a network of edge servers. To accommodate real-world emergency scenarios, we develop a novel threshold mechanism that supports privilege delegation based on user attributes and hash commitments. Extensive security analysis and performance evaluation of VM-PHRs demonstrate that it is scalable, secure, and practical.</div></div>","PeriodicalId":50027,"journal":{"name":"Journal of Systems Architecture","volume":"172 ","pages":"Article 103689"},"PeriodicalIF":4.1,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145927094","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Accelerating language giants: A survey of optimization strategies for LLM inference on hardware platforms 加速语言巨头：硬件平台上LLM推理优化策略的调查

IF 4.1 2区计算机科学 Q1 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Journal of Systems Architecture

Pub Date : 2026-03-01 Epub Date: 2026-01-03 DOI: 10.1016/j.sysarc.2026.103690

Young Chan Kim , Seok Kyu Yoon , Sung Soo Han, Chae Won Park, Jun Oh Park, Jun Ha Ko, Hyun Kim

With the emergence of transformer-based models that have demonstrated remarkable performance in natural language processing tasks, large language models (LLMs) built upon the transformer architecture and trained on massive datasets have achieved outstanding results in various tasks such as translation and summarization. Among these, decoder-only LLMs have garnered significant attention due to their superior few-shot and zero-shot capabilities compared to other architectures. Motivated by their exceptional performance, numerous efforts have been made to deploy decoder-only LLMs on diverse hardware platforms. However, the substantial computational and memory demands during both training and inference pose considerable challenges for resource-constrained hardware. Although efficient architectural designs have been proposed to address these issues, LLM inference continues to require excessive computational and memory resources. Consequently, extensive research has been conducted to compress model components and enhance inference efficiency across different hardware platforms. To further accelerate the inherently repetitive computations of LLMs, a variety of approaches have been introduced, integrating operator-level optimizations within Transformer blocks and system-level optimizations at the granularity of repeated Transformer block execution. This paper surveys recent research on decoder-only LLM inference acceleration, categorizing existing approaches based on optimization levels specific to each hardware platform. Building on this classification, we provide a comprehensive analysis of prior decoder-only LLM acceleration techniques from multiple perspectives.

随着基于变压器的模型在自然语言处理任务中表现出色的出现，基于变压器架构并在海量数据集上训练的大型语言模型（llm）在翻译和摘要等各种任务中取得了优异的成绩。其中，与其他架构相比，仅解码器的llm由于其优越的少射和零射能力而引起了极大的关注。由于其卓越的性能，人们已经做出了许多努力，在不同的硬件平台上部署只有解码器的llm。然而，在训练和推理过程中，大量的计算和内存需求对资源受限的硬件构成了相当大的挑战。尽管已经提出了有效的体系结构设计来解决这些问题，但LLM推理仍然需要过多的计算和内存资源。因此，人们对压缩模型组件和提高不同硬件平台的推理效率进行了广泛的研究。为了进一步加速llm固有的重复计算，已经引入了各种方法，在Transformer块中集成了操作员级优化，并在重复Transformer块执行的粒度上集成了系统级优化。本文综述了最近关于仅解码器的LLM推理加速的研究，并基于特定于每个硬件平台的优化级别对现有方法进行了分类。在此分类的基础上，我们从多个角度对先前的仅解码器的LLM加速技术进行了全面分析。

{"title":"Accelerating language giants: A survey of optimization strategies for LLM inference on hardware platforms","authors":"Young Chan Kim , Seok Kyu Yoon , Sung Soo Han, Chae Won Park, Jun Oh Park, Jun Ha Ko, Hyun Kim","doi":"10.1016/j.sysarc.2026.103690","DOIUrl":"10.1016/j.sysarc.2026.103690","url":null,"abstract":"<div><div>With the emergence of transformer-based models that have demonstrated remarkable performance in natural language processing tasks, large language models (LLMs) built upon the transformer architecture and trained on massive datasets have achieved outstanding results in various tasks such as translation and summarization. Among these, decoder-only LLMs have garnered significant attention due to their superior few-shot and zero-shot capabilities compared to other architectures. Motivated by their exceptional performance, numerous efforts have been made to deploy decoder-only LLMs on diverse hardware platforms. However, the substantial computational and memory demands during both training and inference pose considerable challenges for resource-constrained hardware. Although efficient architectural designs have been proposed to address these issues, LLM inference continues to require excessive computational and memory resources. Consequently, extensive research has been conducted to compress model components and enhance inference efficiency across different hardware platforms. To further accelerate the inherently repetitive computations of LLMs, a variety of approaches have been introduced, integrating operator-level optimizations within Transformer blocks and system-level optimizations at the granularity of repeated Transformer block execution. This paper surveys recent research on decoder-only LLM inference acceleration, categorizing existing approaches based on optimization levels specific to each hardware platform. Building on this classification, we provide a comprehensive analysis of prior decoder-only LLM acceleration techniques from multiple perspectives.</div></div>","PeriodicalId":50027,"journal":{"name":"Journal of Systems Architecture","volume":"172 ","pages":"Article 103690"},"PeriodicalIF":4.1,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145927172","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Thwarting gradient inversion in federated learning via generative shadow mapping defense 通过生成阴影映射防御阻止联邦学习中的梯度反转

IF 4.1 2区计算机科学 Q1 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Journal of Systems Architecture

Pub Date : 2026-03-01 Epub Date: 2025-12-22 DOI: 10.1016/j.sysarc.2025.103671

Hui Zhou , Yuling Chen , Zheng Qin , Xin Deng , Ziyu Peng

Federated learning (FL) has garnered significant attention in the Artificial Intelligence of Things (AIoT) domain. It enables collaborative learning across distributed, privacy-sensitive devices without compromising their local data. However, existing research indicates that adversaries can still reconstruct the raw data by the observed gradient, resulting in a privacy breach. To further strengthen privacy in FL, various defense measures have been proposed, ranging from encryption-based and perturbation-based methods to advanced adaptive strategies. However, nearly all such defenses are applied directly to raw data or gradients, where the private information inherently resides. This intrinsic presence of sensitive data inevitably leaves FL vulnerable to privacy leakage. Thus, a new defense that can “erase” private information is urgently needed. In this paper, we propose Shade, a shadow mapping defense framework against gradient inversion attack using generative models in FL. We implement two instances of manifold defense methods based on a generative adversarial networks and diffusion models, ShadeGAN and ShadeDiff. In particular, we first generate alternative shadow data to involve in model training. Subsequently, we construct surrogate model to replace the raw model, eliminating the memory of the raw model. Finally, an optional gradient protection mechanism is provided, which operates by mapping raw gradients to their shadow counterparts. Extensive experiment demonstrates that our scheme can prevent adversaries from reconstructing raw data, effectively reducing the risk of FL privacy disclosure.

联邦学习（FL）在物联网人工智能（AIoT）领域引起了广泛的关注。它支持跨分布式、隐私敏感设备的协作学习，而不会损害其本地数据。然而，现有研究表明，攻击者仍然可以通过观察到的梯度重建原始数据，从而导致隐私泄露。为了进一步加强FL中的隐私，人们提出了各种防御措施，从基于加密和基于扰动的方法到高级自适应策略。然而，几乎所有此类防御都直接应用于原始数据或梯度，其中包含私有信息。这种敏感数据的固有存在不可避免地使FL容易受到隐私泄露的影响。因此，迫切需要一种可以“擦除”私人信息的新防御手段。在本文中，我们提出了阴影映射防御框架Shade，使用FL中的生成模型来对抗梯度反转攻击。我们实现了两个基于生成对抗网络和扩散模型的流形防御方法实例，ShadeGAN和ShadeDiff。特别是，我们首先生成替代阴影数据来参与模型训练。随后，我们构建代理模型来代替原始模型，消除了原始模型的内存。最后，提供了一个可选的渐变保护机制，它通过将原始渐变映射到对应的阴影渐变来操作。大量实验表明，我们的方案可以防止对手重构原始数据，有效降低FL隐私泄露的风险。

{"title":"Thwarting gradient inversion in federated learning via generative shadow mapping defense","authors":"Hui Zhou , Yuling Chen , Zheng Qin , Xin Deng , Ziyu Peng","doi":"10.1016/j.sysarc.2025.103671","DOIUrl":"10.1016/j.sysarc.2025.103671","url":null,"abstract":"<div><div>Federated learning (FL) has garnered significant attention in the Artificial Intelligence of Things (AIoT) domain. It enables collaborative learning across distributed, privacy-sensitive devices without compromising their local data. However, existing research indicates that adversaries can still reconstruct the raw data by the observed gradient, resulting in a privacy breach. To further strengthen privacy in FL, various defense measures have been proposed, ranging from encryption-based and perturbation-based methods to advanced adaptive strategies. However, nearly all such defenses are applied directly to raw data or gradients, where the private information inherently resides. This intrinsic presence of sensitive data inevitably leaves FL vulnerable to privacy leakage. Thus, a new defense that can <em>“erase”</em> private information is urgently needed. In this paper, we propose <em>Shade</em>, a shadow mapping defense framework against gradient inversion attack using generative models in FL. We implement two instances of manifold defense methods based on a generative adversarial networks and diffusion models, <em>ShadeGAN</em> and <em>ShadeDiff</em>. In particular, we first generate alternative shadow data to involve in model training. Subsequently, we construct surrogate model to replace the raw model, eliminating the memory of the raw model. Finally, an optional gradient protection mechanism is provided, which operates by mapping raw gradients to their shadow counterparts. Extensive experiment demonstrates that our scheme can prevent adversaries from reconstructing raw data, effectively reducing the risk of FL privacy disclosure.</div></div>","PeriodicalId":50027,"journal":{"name":"Journal of Systems Architecture","volume":"172 ","pages":"Article 103671"},"PeriodicalIF":4.1,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145841939","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

DeSpa: Heterogeneous multi-core accelerators for energy-efficient dense and sparse computation at the tile level in Deep Neural Networks 基于异构多核加速器的深度神经网络层级高效密集稀疏计算

IF 4.1 2区计算机科学 Q1 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Journal of Systems Architecture

Pub Date : 2026-03-01 Epub Date: 2025-12-12 DOI: 10.1016/j.sysarc.2025.103650

Hyungjun Jang , Dongho Ha , Hyunwuk Lee , Won Woo Ro

The rapid evolution of Deep Neural Networks (DNNs) has driven significant advances in Domain-Specific Accelerators (DSAs). However, efficiently exploiting DSAs across diverse workloads remains challenging because complementary techniques—from sparsity-aware computation to system-level innovations such as multi-core architectures—have progressed independently. Our analysis reveals pronounced tile-level sparsity variations within the DNNs, which cause efficiency fluctuations on homogeneous accelerators built solely from dense or sparsity-oriented cores. To address this challenge, we present DeSpa, a novel heterogeneous multi-core accelerator architecture that integrates both dense and sparse cores to dynamically adapt to tile-level sparsity variations. DeSpa is paired with a heterogeneity-aware scheduler that employs a tile-stealing mechanism to maximize core utilization and minimize idle time. Compared to a homogeneous sparse multi-core baseline, DeSpa reduces energy consumption by 33% and improves energy-delay product (EDP) by 14%, albeit at the cost of a 35% latency increase. Relative to a homogeneous dense baseline, it reduces EDP by 44%, cuts energy consumption by 42%, and delivers a

1.34 \times

speed-up.

深度神经网络（dnn）的快速发展推动了特定领域加速器（dsa）的重大进展。然而，跨不同工作负载有效地利用dsa仍然具有挑战性，因为互补技术——从稀疏感知计算到系统级创新（如多核体系结构）——已经独立发展。我们的分析揭示了dnn内明显的瓷砖级稀疏性变化，这导致仅由密集或稀疏性导向核心构建的均匀加速器的效率波动。为了应对这一挑战，我们提出了DeSpa，这是一种新型的异构多核加速器架构，它集成了密集核和稀疏核，以动态适应瓷砖级稀疏度变化。DeSpa与一个异构感知调度器配合使用，该调度器使用一种磁片窃取机制来最大化核心利用率并最小化空闲时间。与同质稀疏多核基线相比，DeSpa减少了33%的能耗，并将能量延迟产品（EDP）提高了14%，但代价是延迟增加了35%。相对于均匀的密集基线，它可以降低44%的EDP，降低42%的能耗，并提供1.34倍的加速。

{"title":"DeSpa: Heterogeneous multi-core accelerators for energy-efficient dense and sparse computation at the tile level in Deep Neural Networks","authors":"Hyungjun Jang , Dongho Ha , Hyunwuk Lee , Won Woo Ro","doi":"10.1016/j.sysarc.2025.103650","DOIUrl":"10.1016/j.sysarc.2025.103650","url":null,"abstract":"<div><div>The rapid evolution of Deep Neural Networks (DNNs) has driven significant advances in Domain-Specific Accelerators (DSAs). However, efficiently exploiting DSAs across diverse workloads remains challenging because complementary techniques—from sparsity-aware computation to system-level innovations such as multi-core architectures—have progressed independently. Our analysis reveals pronounced tile-level sparsity variations within the DNNs, which cause efficiency fluctuations on homogeneous accelerators built solely from dense or sparsity-oriented cores. To address this challenge, we present DeSpa, a novel heterogeneous multi-core accelerator architecture that integrates both dense and sparse cores to dynamically adapt to tile-level sparsity variations. DeSpa is paired with a heterogeneity-aware scheduler that employs a tile-stealing mechanism to maximize core utilization and minimize idle time. Compared to a homogeneous sparse multi-core baseline, DeSpa reduces energy consumption by 33% and improves energy-delay product (EDP) by 14%, albeit at the cost of a 35% latency increase. Relative to a homogeneous dense baseline, it reduces EDP by 44%, cuts energy consumption by 42%, and delivers a <span><math><mrow><mn>1</mn><mo>.</mo><mn>34</mn><mo>×</mo></mrow></math></span> speed-up.</div></div>","PeriodicalId":50027,"journal":{"name":"Journal of Systems Architecture","volume":"172 ","pages":"Article 103650"},"PeriodicalIF":4.1,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145792174","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

ARAS: Adaptive low-cost ReRAM-based accelerator for DNNs ARAS：用于深度神经网络的自适应低成本rram加速器

IF 4.1 2区计算机科学 Q1 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Journal of Systems Architecture

Pub Date : 2026-03-01 Epub Date: 2025-12-20 DOI: 10.1016/j.sysarc.2025.103668

Mohammad Sabri, Marc Riera, Antonio González

Processing Using Memory (PUM) accelerators have the potential to perform Deep Neural Network (DNN) inference by using arrays of memory cells as computation engines. Among various memory technologies, ReRAM crossbars show promising performance in computing dot-product operations in the analog domain. Nevertheless, the expensive writing procedure of ReRAM cells has led researchers to design accelerators whose crossbars have enough capacity to store the full DNN. Given the tremendous and continuous increase in DNN model sizes, this approach is unfeasible for some networks, or inefficient due to the huge hardware requirements. Those accelerators lack the flexibility to adapt to any given DNN model, facing an adaptability challenge.

To address this issue we introduce ARAS, a cost-effective ReRAM-based accelerator that employs an offline scheduler to adapt different DNNs to the resource-limited hardware. ARAS also overlaps the computation of a layer with the weight writing of several layers to mitigate the high writing latency of ReRAM. Furthermore, ARAS introduces three optimizations aimed at reducing the energy overheads of writing in ReRAM. Our key optimization capitalizes on the observation that DNN weights can be re-encoded to augment their similarity between layers, increasing the amount of bitwise values that are equal or similar when overwriting ReRAM cells and, hence, reducing the amount of energy required to update the cells. Overall, ARAS greatly reduces the ReRAM writing activity. We evaluate ARAS on a popular set of DNNs. ARAS provides up to

2.2 \times

speedup and 45% energy savings over a baseline PUM accelerator without any optimization. Compared to a TPU-like accelerator, ARAS provides up to

1.5 \times

speedup and 62% energy savings.

使用内存处理（PUM）加速器有可能通过使用存储单元阵列作为计算引擎来执行深度神经网络（DNN）推理。在各种存储技术中，ReRAM交叉棒在模拟域计算点积运算方面表现出良好的性能。然而，昂贵的ReRAM细胞写入过程促使研究人员设计出具有足够容量存储完整深度神经网络的加速器。考虑到DNN模型规模的巨大和持续增长，这种方法对于某些网络是不可行的，或者由于巨大的硬件要求而效率低下。这些加速器缺乏适应任何给定DNN模型的灵活性，面临适应性挑战。为了解决这个问题，我们引入了ARAS，这是一种经济高效的基于reram的加速器，它采用离线调度器来适应不同的dnn以适应资源有限的硬件。ARAS还将一个层的计算与多个层的权重写入重叠，以减轻ReRAM的高写入延迟。此外，ARAS引入了三种优化，旨在减少在ReRAM中写入的能量开销。我们的关键优化利用了DNN权重可以被重新编码以增加层之间的相似性的观察结果，增加了覆盖ReRAM单元时相等或相似的位值的数量，从而减少了更新单元所需的能量。总的来说，ARAS大大减少了ReRAM写入活动。我们在一组流行的dnn上评估ARAS。在没有任何优化的情况下，与基准PUM加速器相比，ARAS提供高达2.2倍的加速和45%的节能。与类似tpu的加速器相比，ARAS提供高达1.5倍的加速和62%的节能。

{"title":"ARAS: Adaptive low-cost ReRAM-based accelerator for DNNs","authors":"Mohammad Sabri, Marc Riera, Antonio González","doi":"10.1016/j.sysarc.2025.103668","DOIUrl":"10.1016/j.sysarc.2025.103668","url":null,"abstract":"<div><div>Processing Using Memory (PUM) accelerators have the potential to perform Deep Neural Network (DNN) inference by using arrays of memory cells as computation engines. Among various memory technologies, ReRAM crossbars show promising performance in computing dot-product operations in the analog domain. Nevertheless, the expensive writing procedure of ReRAM cells has led researchers to design accelerators whose crossbars have enough capacity to store the full DNN. Given the tremendous and continuous increase in DNN model sizes, this approach is unfeasible for some networks, or inefficient due to the huge hardware requirements. Those accelerators lack the flexibility to adapt to any given DNN model, facing an <em>adaptability</em> challenge.</div><div>To address this issue we introduce ARAS, a cost-effective ReRAM-based accelerator that employs an offline scheduler to adapt different DNNs to the resource-limited hardware. ARAS also overlaps the computation of a layer with the weight writing of several layers to mitigate the high writing latency of ReRAM. Furthermore, ARAS introduces three optimizations aimed at reducing the energy overheads of writing in ReRAM. Our key optimization capitalizes on the observation that DNN weights can be re-encoded to augment their similarity between layers, increasing the amount of bitwise values that are equal or similar when overwriting ReRAM cells and, hence, reducing the amount of energy required to update the cells. Overall, ARAS greatly reduces the ReRAM writing activity. We evaluate ARAS on a popular set of DNNs. ARAS provides up to <span><math><mrow><mn>2</mn><mo>.</mo><mn>2</mn><mo>×</mo></mrow></math></span> speedup and 45% energy savings over a baseline PUM accelerator without any optimization. Compared to a TPU-like accelerator, ARAS provides up to <span><math><mrow><mn>1</mn><mo>.</mo><mn>5</mn><mo>×</mo></mrow></math></span> speedup and 62% energy savings.</div></div>","PeriodicalId":50027,"journal":{"name":"Journal of Systems Architecture","volume":"172 ","pages":"Article 103668"},"PeriodicalIF":4.1,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145885131","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Efficient column-wise N:M pruning on RISC-V CPU RISC-V CPU上高效的按列N:M修剪

IF 4.1 2区计算机科学 Q1 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Journal of Systems Architecture

Pub Date : 2026-03-01 Epub Date: 2025-12-26 DOI: 10.1016/j.sysarc.2025.103667

Chi-Wei Chu, Ding-Yong Hong, Jan-Jan Wu

In deep learning frameworks, weight pruning is a widely used technique for improving computational efficiency by reducing the size of large models. This is especially critical for convolutional operators, which often act as performance bottlenecks in convolutional neural networks (CNNs). However, the effectiveness of pruning heavily depends on how it is implemented, as different methods can significantly impact both computational performance and memory footprint. In this work, we propose a column-wise N:M pruning strategy applied at the tile level and modify XNNPACK to enable efficient execution of pruned models on the RISC-V vector architecture. Additionally, we propose fusing the operations of im2col and data packing to minimize redundant memory accesses and memory overhead. To further optimize performance, we incorporate AITemplate’s profiling technique to identify the optimal implementation for each convolutional operator. Our proposed approach effectively increases ResNet inference throughput by as much as

4 \times

, and preserves ImageNet top-1 accuracy within 2.1% of the dense baseline. The code of our framework is publicly available at https://github.com/wewe5215/AI_template_RVV_backend

在深度学习框架中，权值修剪是一种广泛使用的技术，通过减少大型模型的大小来提高计算效率。这对于卷积算子来说尤其重要，卷积算子经常成为卷积神经网络（cnn）的性能瓶颈。然而，剪枝的有效性在很大程度上取决于它是如何实现的，因为不同的方法会显著影响计算性能和内存占用。在这项工作中，我们提出了一种在tile级别应用的列式N:M修剪策略，并修改了XNNPACK，以便在RISC-V矢量架构上有效地执行修剪模型。此外，我们建议融合im2col和数据打包的操作，以尽量减少冗余内存访问和内存开销。为了进一步优化性能，我们结合了AITemplate的分析技术来确定每个卷积算子的最佳实现。我们提出的方法有效地将ResNet推理吞吐量提高了4倍，并将ImageNet top-1精度保持在密集基线的2.1%以内。我们的框架的代码可以在https://github.com/wewe5215/AI_template_RVV_backend上公开获得

{"title":"Efficient column-wise N:M pruning on RISC-V CPU","authors":"Chi-Wei Chu, Ding-Yong Hong, Jan-Jan Wu","doi":"10.1016/j.sysarc.2025.103667","DOIUrl":"10.1016/j.sysarc.2025.103667","url":null,"abstract":"<div><div>In deep learning frameworks, weight pruning is a widely used technique for improving computational efficiency by reducing the size of large models. This is especially critical for convolutional operators, which often act as performance bottlenecks in convolutional neural networks (CNNs). However, the effectiveness of pruning heavily depends on how it is implemented, as different methods can significantly impact both computational performance and memory footprint. In this work, we propose a column-wise N:M pruning strategy applied at the tile level and modify XNNPACK to enable efficient execution of pruned models on the RISC-V vector architecture. Additionally, we propose fusing the operations of im2col and data packing to minimize redundant memory accesses and memory overhead. To further optimize performance, we incorporate AITemplate’s profiling technique to identify the optimal implementation for each convolutional operator. Our proposed approach effectively increases ResNet inference throughput by as much as <span><math><mrow><mn>4</mn><mo>×</mo></mrow></math></span>, and preserves ImageNet top-1 accuracy within 2.1% of the dense baseline. The code of our framework is publicly available at <span><span>https://github.com/wewe5215/AI_template_RVV_backend</span><svg><path></path></svg></span></div></div>","PeriodicalId":50027,"journal":{"name":"Journal of Systems Architecture","volume":"172 ","pages":"Article 103667"},"PeriodicalIF":4.1,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145885132","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

An efficient privacy-preserving transformer inference scheme for cloud-based intelligent decision-making in AIoT 基于云的AIoT智能决策中一种高效的保隐私变压器推理方案

IF 4.1 2区计算机科学 Q1 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Journal of Systems Architecture

Pub Date : 2026-03-01 Epub Date: 2026-01-02 DOI: 10.1016/j.sysarc.2026.103687

Mingshun Luo , Haolei He , Wenti Yang , Shuai Yuan , Zhitao Guan

The Artificial Intelligence of Things (AIoT) is transforming modern society by combining the data-collection capabilities of IoT devices with the inference power of cloud-based large language models (LLMs). However, transmitting sensitive data to the cloud for intelligent decision-making raises significant privacy concerns. Cryptographic techniques such as homomorphic encryption (HE) and secure multi-party computation (MPC) provide promising solutions for privacy-preserving inference. However, existing schemes primarily target small-scale models and are inefficient when applied to Transformer-based LLMs, which involve large-scale matrix multiplications and complex non-linear functions, and deep model architectures. To address these challenges, we propose an efficient privacy-preserving Transformer inference scheme for cloud-based AIoT scenarios. Our framework integrates HE and MPC to ensure data confidentiality while minimizing computational and communication overhead. We design a fast HE-based matrix multiplication protocol using an offline-online collaborative pipeline and single instruction multiple data (SIMD)-based packing rules. Furthermore, we develop an accurate and efficient MPC-based non-linear function evaluation protocol using optimized piecewise polynomial approximation and integer-fraction decomposition. Experimental results show that our approach achieves 8.3

\times

–91.6

\times

faster in matrix multiplication, 1.4

\times

–19

\times

faster in non-linear function evaluation, and 3.5

\times

–137.9

\times

reduction in communication overhead with the LAN network, while maintaining lossless accuracy, thus enabling secure and scalable intelligent decision-making in AIoT environments.

物联网人工智能（AIoT）通过将物联网设备的数据收集能力与基于云的大型语言模型（llm）的推理能力相结合，正在改变现代社会。然而，将敏感数据传输到云端以进行智能决策引发了严重的隐私问题。同态加密（HE）和安全多方计算（MPC）等密码学技术为隐私保护推理提供了有前途的解决方案。然而，现有方案主要针对小规模模型，并且在应用于基于变压器的llm时效率低下，这涉及大规模矩阵乘法和复杂的非线性函数，以及深度模型架构。为了解决这些挑战，我们为基于云的AIoT场景提出了一种有效的隐私保护Transformer推理方案。我们的框架集成了HE和MPC，以确保数据机密性，同时最大限度地减少计算和通信开销。我们使用离线-在线协同管道和基于单指令多数据（SIMD）的打包规则设计了一个基于he的快速矩阵乘法协议。此外，我们开发了一种精确和高效的基于mpc的非线性函数评估协议，该协议采用优化的分段多项式近似和积分分数分解。实验结果表明，我们的方法在保持无损精度的同时，实现了8.3×-91.6×更快的矩阵乘法速度，1.4×-19×更快的非线性函数求值速度，3.5×-137.9×减少了与局域网网络的通信开销，从而实现了AIoT环境中安全且可扩展的智能决策。

{"title":"An efficient privacy-preserving transformer inference scheme for cloud-based intelligent decision-making in AIoT","authors":"Mingshun Luo , Haolei He , Wenti Yang , Shuai Yuan , Zhitao Guan","doi":"10.1016/j.sysarc.2026.103687","DOIUrl":"10.1016/j.sysarc.2026.103687","url":null,"abstract":"<div><div>The Artificial Intelligence of Things (AIoT) is transforming modern society by combining the data-collection capabilities of IoT devices with the inference power of cloud-based large language models (LLMs). However, transmitting sensitive data to the cloud for intelligent decision-making raises significant privacy concerns. Cryptographic techniques such as homomorphic encryption (HE) and secure multi-party computation (MPC) provide promising solutions for privacy-preserving inference. However, existing schemes primarily target small-scale models and are inefficient when applied to Transformer-based LLMs, which involve large-scale matrix multiplications and complex non-linear functions, and deep model architectures. To address these challenges, we propose an efficient privacy-preserving Transformer inference scheme for cloud-based AIoT scenarios. Our framework integrates HE and MPC to ensure data confidentiality while minimizing computational and communication overhead. We design a fast HE-based matrix multiplication protocol using an offline-online collaborative pipeline and single instruction multiple data (SIMD)-based packing rules. Furthermore, we develop an accurate and efficient MPC-based non-linear function evaluation protocol using optimized piecewise polynomial approximation and integer-fraction decomposition. Experimental results show that our approach achieves 8.3<span><math><mo>×</mo></math></span>–91.6<span><math><mo>×</mo></math></span> faster in matrix multiplication, 1.4<span><math><mo>×</mo></math></span>–19<span><math><mo>×</mo></math></span> faster in non-linear function evaluation, and 3.5<span><math><mo>×</mo></math></span>–137.9<span><math><mo>×</mo></math></span> reduction in communication overhead with the LAN network, while maintaining lossless accuracy, thus enabling secure and scalable intelligent decision-making in AIoT environments.</div></div>","PeriodicalId":50027,"journal":{"name":"Journal of Systems Architecture","volume":"172 ","pages":"Article 103687"},"PeriodicalIF":4.1,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145927095","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Group theory-based differential evolution algorithm for efficient DAG scheduling on heterogeneous clustered multi-core system 基于群理论的异构集群多核系统DAG高效调度差分进化算法

IF 4.1 2区计算机科学 Q1 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Journal of Systems Architecture

Pub Date : 2026-03-01 Epub Date: 2026-01-05 DOI: 10.1016/j.sysarc.2026.103695

Yaodong Guo, Shuangshuang Chang, Dong Ji, Shiyue Qin, Te Xu

Efficient parallel application scheduling algorithms are crucial for optimizing performance on heterogeneous clustered multi-core systems. The primary objective of scheduling is to reduce the makespan of parallel applications, typically represented as Directed Acyclic Graphs (DAGs). This paper introduces a Group Theory-based Differential Evolution (GTDE¹) algorithm to address the NP-complete DAG scheduling problem, to minimize makespan and computation time. The GTDE algorithm leverages group theory to explore the inherent symmetry in system architectures, enabling the classification of scheduling schemes and thus reducing redundant computations while maintaining population diversity. To further enhance performance, the algorithm employs an Opposition-Based Learning (OBL) strategy to improve the initial population and integrates a hybrid mutation strategy for more efficient exploration of the solution space. Experimental results demonstrate that the GTDE algorithm consistently outperforms state-of-the-art DAG scheduling algorithms in terms of performance metrics, such as makespan and computation time, with average improvements of 36% and 73%, respectively, achieving superior performance across various scenarios.

高效的并行应用程序调度算法对于优化异构集群多核系统的性能至关重要。调度的主要目标是减少并行应用程序的最大运行时间，通常表示为有向无环图（dag）。本文提出了一种基于群论的差分进化算法（GTDE1）来解决np完全DAG调度问题，以最小化完工时间和计算时间。GTDE算法利用群理论来探索系统架构中固有的对称性，实现调度方案的分类，从而在保持种群多样性的同时减少冗余计算。为了进一步提高性能，该算法采用了基于对立的学习（OBL）策略来改进初始种群，并集成了混合突变策略来更有效地探索解空间。实验结果表明，GTDE算法在makespan和计算时间等性能指标上始终优于最先进的DAG调度算法，平均分别提高36%和73%，在各种场景下都具有卓越的性能。

{"title":"Group theory-based differential evolution algorithm for efficient DAG scheduling on heterogeneous clustered multi-core system","authors":"Yaodong Guo, Shuangshuang Chang, Dong Ji, Shiyue Qin, Te Xu","doi":"10.1016/j.sysarc.2026.103695","DOIUrl":"10.1016/j.sysarc.2026.103695","url":null,"abstract":"<div><div>Efficient parallel application scheduling algorithms are crucial for optimizing performance on heterogeneous clustered multi-core systems. The primary objective of scheduling is to reduce the makespan of parallel applications, typically represented as Directed Acyclic Graphs (DAGs). This paper introduces a Group Theory-based Differential Evolution (GTDE<span><span><sup>1</sup></span></span>) algorithm to address the NP-complete DAG scheduling problem, to minimize makespan and computation time. The GTDE algorithm leverages group theory to explore the inherent symmetry in system architectures, enabling the classification of scheduling schemes and thus reducing redundant computations while maintaining population diversity. To further enhance performance, the algorithm employs an Opposition-Based Learning (OBL) strategy to improve the initial population and integrates a hybrid mutation strategy for more efficient exploration of the solution space. Experimental results demonstrate that the GTDE algorithm consistently outperforms state-of-the-art DAG scheduling algorithms in terms of performance metrics, such as makespan and computation time, with average improvements of 36% and 73%, respectively, achieving superior performance across various scenarios.</div></div>","PeriodicalId":50027,"journal":{"name":"Journal of Systems Architecture","volume":"172 ","pages":"Article 103695"},"PeriodicalIF":4.1,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145927173","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

MxGPU: Efficient and safe communication between GPGPU applications in an OS-controlled GPGPU multiplexing environment MxGPU：在操作系统控制的GPGPU多路复用环境中，实现GPGPU应用间高效、安全的通信

IF 4.1 2区计算机科学 Q1 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Journal of Systems Architecture

Pub Date : 2026-03-01 Epub Date: 2025-12-24 DOI: 10.1016/j.sysarc.2025.103669

Marcel Lütke Dreimann, Olaf Spinczyk

With the growing demand for artificial intelligence and other data-intensive applications, the demand for graphics processing units (GPUs) has also increased. Even though there are many approaches on multiplexing GPUs, none of the approaches known to us enable the operating system to coherently integrate GPU resources alongside CPU resources into a holistic resource management. Due to the history of GPUs, GPU drivers are still a large, isolated part within the driver stack of operating systems. This paper aims to conduct a case study on how a multiplexing solution for GPGPUs could look like, where the OS is able to define scheduling policies for GPGPU tasks and manage GPU memory. OS-controlled GPU memory management can especially be helpful for efficient and safe communication between GPGPU applications. We will discuss and evaluate the architecture of MxGPU, which offers software-based multiplexing of integrated Intel GPUs. MxGPU has a tiny code base, which is a precondition for formal verification approaches and usage in safety-critical environments. Experiments with our prototype show that MxGPU can grant the operating system control over GPU resources while allowing more GPU sessions. Furthermore, MxGPU allows for execution of GPGPU tasks with less latency compared to Linux and enables efficient and safe communication between GPU applications.

随着对人工智能和其他数据密集型应用的需求不断增长，对图形处理单元（gpu）的需求也在增加。尽管在多路GPU上有很多方法，但我们所知道的方法都不能使操作系统将GPU资源和CPU资源一致地集成到一个整体的资源管理中。由于GPU的历史，GPU驱动程序仍然是操作系统驱动程序堆栈中一个很大的、孤立的部分。本文旨在对GPGPU的多路复用解决方案进行案例研究，其中操作系统能够为GPGPU任务定义调度策略并管理GPU内存。操作系统控制的GPU内存管理对于GPGPU应用程序之间高效和安全的通信特别有帮助。我们将讨论和评估MxGPU的架构，它提供基于软件的集成英特尔gpu的多路复用。MxGPU的代码库很小，这是正式验证方法和在安全关键环境中使用的先决条件。我们的原型实验表明，MxGPU可以在允许更多GPU会话的同时授予操作系统对GPU资源的控制。此外，与Linux相比，MxGPU允许以更少的延迟执行GPGPU任务，并在GPU应用程序之间实现高效和安全的通信。

{"title":"MxGPU: Efficient and safe communication between GPGPU applications in an OS-controlled GPGPU multiplexing environment","authors":"Marcel Lütke Dreimann, Olaf Spinczyk","doi":"10.1016/j.sysarc.2025.103669","DOIUrl":"10.1016/j.sysarc.2025.103669","url":null,"abstract":"<div><div>With the growing demand for artificial intelligence and other data-intensive applications, the demand for graphics processing units (GPUs) has also increased. Even though there are many approaches on multiplexing GPUs, none of the approaches known to us enable the operating system to coherently integrate GPU resources alongside CPU resources into a holistic resource management. Due to the history of GPUs, GPU drivers are still a large, isolated part within the driver stack of operating systems. This paper aims to conduct a case study on how a multiplexing solution for GPGPUs could look like, where the OS is able to define scheduling policies for GPGPU tasks and manage GPU memory. OS-controlled GPU memory management can especially be helpful for efficient and safe communication between GPGPU applications. We will discuss and evaluate the architecture of <span>MxGPU</span>, which offers software-based multiplexing of integrated Intel GPUs. <span>MxGPU</span> has a tiny code base, which is a precondition for formal verification approaches and usage in safety-critical environments. Experiments with our prototype show that <span>MxGPU</span> can grant the operating system control over GPU resources while allowing more GPU sessions. Furthermore, <span>MxGPU</span> allows for execution of GPGPU tasks with less latency compared to Linux and enables efficient and safe communication between GPU applications.</div></div>","PeriodicalId":50027,"journal":{"name":"Journal of Systems Architecture","volume":"172 ","pages":"Article 103669"},"PeriodicalIF":4.1,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145841942","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Towards privacy preservation in smart grids via controlled redactable signatures 通过可控可读签名实现智能电网的隐私保护

IF 4.1 2区计算机科学 Q1 COMPUTER SCIENCE, HARDWARE & ARCHITECTURE

Journal of Systems Architecture

Pub Date : 2026-03-01 Epub Date: 2026-01-02 DOI: 10.1016/j.sysarc.2026.103688

Siyuan Shen , Xiaoying Jia , Min Luo , Zhiyan Xu , Zhichao Zhou

The smart grid provides a flexible interactive platform for energy stakeholders by sharing energy usage data to enhance management efficiency and service accuracy. However, such data are often highly sensitive and vulnerable to eavesdropping and tampering during transmission. Ensuring data authenticity, integrity and users’ privacy is therefore critical. Redactable signatures have emerged as a promising cryptographic primitive to address these concerns. Nonetheless, most existing redactable signature schemes lack fine-grained control over the redaction process, making them susceptible to unauthorized or malicious modifications. To address this issue, we propose an identity-based Controlled Redactable Signature Scheme (CRSS), enabling users to selectively disclose information under controlled conditions without revealing private information. We define a formal security model and prove that the proposed scheme achieves unforgeability, redaction controllability, privacy, and transparency. Furthermore, theoretical analysis and experimental evaluation demonstrate that our scheme offers superior efficiency and practicality compared to existing approaches.

智能电网通过共享能源使用数据，为能源利益相关者提供灵活的互动平台，提高管理效率和服务准确性。然而，这些数据往往高度敏感，在传输过程中容易被窃听和篡改。因此，确保数据的真实性、完整性和用户隐私至关重要。可读签名已经成为解决这些问题的一种很有前途的加密原语。尽管如此，大多数现有的可重写签名方案缺乏对编校过程的细粒度控制，使它们容易受到未经授权或恶意修改的影响。为了解决这个问题，我们提出了一种基于身份的受控可读签名方案（CRSS），使用户能够在受控条件下选择性地披露信息而不会泄露私人信息。我们定义了一个形式化的安全模型，并证明了该方案具有不可伪造性、编校可控性、隐私性和透明性。理论分析和实验评价表明，与现有方法相比，该方案具有更高的效率和实用性。

{"title":"Towards privacy preservation in smart grids via controlled redactable signatures","authors":"Siyuan Shen , Xiaoying Jia , Min Luo , Zhiyan Xu , Zhichao Zhou","doi":"10.1016/j.sysarc.2026.103688","DOIUrl":"10.1016/j.sysarc.2026.103688","url":null,"abstract":"<div><div>The smart grid provides a flexible interactive platform for energy stakeholders by sharing energy usage data to enhance management efficiency and service accuracy. However, such data are often highly sensitive and vulnerable to eavesdropping and tampering during transmission. Ensuring data authenticity, integrity and users’ privacy is therefore critical. Redactable signatures have emerged as a promising cryptographic primitive to address these concerns. Nonetheless, most existing redactable signature schemes lack fine-grained control over the redaction process, making them susceptible to unauthorized or malicious modifications. To address this issue, we propose an identity-based Controlled Redactable Signature Scheme (CRSS), enabling users to selectively disclose information under controlled conditions without revealing private information. We define a formal security model and prove that the proposed scheme achieves unforgeability, redaction controllability, privacy, and transparency. Furthermore, theoretical analysis and experimental evaluation demonstrate that our scheme offers superior efficiency and practicality compared to existing approaches.</div></div>","PeriodicalId":50027,"journal":{"name":"Journal of Systems Architecture","volume":"172 ","pages":"Article 103688"},"PeriodicalIF":4.1,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145927096","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0