Request Size Research Articles

Efficient resource allocation for IoT applications in mobile edge computing via dynamic request scheduling optimization

This study investigates the challenges of task scheduling and resource allocation in an ultra-dense edge cloud (UDEC) network incorporating micro (macro) base stations and different equipment users (UE) under 5G technology. By utilizing multiple access interference (MAI), we enable multiple access capabilities to the UEs under the Non-Orthogonal Multiple Access (NOMA) protocol. To accommodate the dynamic nature of user tasks, we propose a two-level scheduling framework. At the upper level, the non-convex, non-linear power allocation problem for mobile users is formulated and solved using the Inertia Weight Particle Swarm Optimization (IW-PSO) algorithm to minimize the transmission energy consumption. Simultaneously, at the lower level, the joint task offloading and resource allocation problem is formulated as MINP and solved with the binary Particle Swarm Optimization (PSO) algorithm to find an optimal schedule that maximizes the response rate and the overall welfare of the system while minimizing the energy consumption. In this study, various sensitivity analyses are conducted for different parameters, including the number of mobile users, channel power, and request profile parameters, including workload and request size. Comparative analyses with other methods from recent literature consistently reveal the comparable performance of the proposed framework and confirm its effectiveness in addressing the challenges of dynamic task scheduling.

Rejection probability reduction in queueing systems with limited queue using size based prioritization and artificial intelligence

The article is dedicated to enhancing the efficiency of request processing in multi-channel mass service systems with limited queue and refusals in case of overflow. The subject of this article is the methods and approaches to optimizing the processing of request flows. The aim of the work is to propose a new approach to prioritizing and balancing request categories to reduce the probability of refusals. The tasks of the article include: formulating the studied mass service system; determining the source of optimization; describing the method of dividing the overall request flow into categories; listing approaches to determining the request sizes; defining the algorithm for processing requests using size-based prioritization; proposing solutions to load assessment and priority balancing problems using artificial intelligence. The following methods are used: mass service theory, UML diagrams, artificial intelligence. The following results were obtained: a method to reduce the probability of refusals in multi-channel mass service systems with a limited queue by increasing the priority of smaller requests was proposed; methods for assessing request complexity and dividing the overall request flow into categories according to their complexity were proposed; an approach to balancing priorities using artificial intelligence was proposed.

PMRNA: Parameter matching of realtime and non‐realtime applications for resource provisioning in fog‐integrated cloud

SummaryFog computing, an emerging technology, extends Cloud computing services to the network's edge in the proximity of the application request. This extension yields improvement in Bandwidth (BW) utilization, faster responses to Real‐Time (RT) and Internet of Things (IoT) requests, and the provision of the heterogeneous resource services. While extensive work has been conducted on resource allocation for RT and Non‐Real‐Time (NRT) requests separately in Fog as well as Cloud computing, there is limited focus on resource provisioning for mixed RT and NRT requests in the Fog‐integrated Cloud (FiC) environment. Moreover, the majority of the existing provisioning methods primarily consider parameters from the system's perspective, overlooking crucial user aspects such as deadline and request size. To address the gap, this work introduces a resource provisioning method named “Parameter Matching of Realtime and Non‐Realtime Applications (PMRNA),” which considers user parameters and resource information, gathered by the broker. The performance evaluation of the proposed model is done in CloudSim, using various combinations of RT and NRT requests along with diverse Fog and Cloud resource configurations. Evaluation metric includes average Execution Time (ET), average Waiting Time (WT), average Turn Around Time (TAT), and resource utilization. The experimental results demonstrate a significant reduction in both average WT and average TAT for the diverse pool of RT and NRT requests in the FiC compared to the Cloud‐only environment.

Reinforcement learning-based dynamic load balancing in edge computing networks

Edge computing (EC) has emerged as a paradigm aimed at reducing data transmission latency by bringing computing resources closer to users. However, the limited scale and constrained processing power of EC pose challenges in matching the resource availability of larger cloud networks. Load balancing (LB) algorithms play a crucial role in distributing workload among edge servers and minimizing user latency. This paper presents a novel set of distributed LB algorithms that leverage machine learning techniques to overcome the three limitations of our previous LB algorithm, EVBLB: (i) its reliance on static time intervals for execution, (ii) the need for comprehensive information about all server resources and queued requests for neighbor selection, and (iii) the use of a central coordinator to dispatch incoming user requests over edge servers. To offer increased control, custom configuration, and scalability for LB on edge servers, we propose three efficient algorithms: Q-learning (QL), multi-armed bandit (MAB), and gradient bandit (GB) algorithms. The QL algorithm predicts the subsequent execution time of the EVBLB algorithm by incorporating rewards obtained from previous executions, thereby improving performance across various metrics. The MAB and GB algorithms prioritize near-optimal neighbor node servers while considering dynamic changes in request rate, request size, and edge server resources. Through simulations, we evaluate and compare the algorithms in terms of network throughput, average user response time, and a novel LB metric for workload distribution across edge servers.

PMAlloc: A Holistic Approach to Improving Persistent Memory Allocation

Persistent memory allocation is a fundamental building block for developing high-performance and in-memory applications. Existing persistent memory allocators suffer from many performance issues. First, they may introduce repeated cache line flushes and small random accesses in persistent memory for their poor heap metadata management. Second, they use static slab segregation resulting in a dramatic increase in memory consumption when allocation request size is changed. Third, they are not aware of NUMA effect, leading to remote persistent memory accesses in memory allocation and deallocation processes. In this paper, we design a novel allocator, named PMAlloc, to solve the above issues simultaneously. (1) PMAlloc eliminates cache line reflushes by mapping contiguous data blocks in slabs to interleaved metadata entries stored in different cache lines. (2) It writes small metadata units to a persistent bookkeeping log in a sequential pattern to remove random heap metadata accesses in persistent memory. (3) Instead of using static slab segregation, it supports slab morphing, which allows slabs to be transformed between size classes to significantly improve slab usage. (4) It uses a local-first allocation policy to avoid allocating remote memory blocks. And it supports a two-phase deallocation mechanism including recording and synchronization to minimize the number of remote memory access in the deallocation. PMAlloc is complementary to the existing consistency models. Results on 6 benchmarks demonstrate that PMAlloc improves the performance of state-of-the-art persistent memory allocators by up to 6.4x and 57x for small and large allocations, respectively. PMAlloc with NUMA optimizations brings a 2.9x speedup in multi-socket evaluation and is up to 36x faster than other persistent memory allocators. Using PMAlloc reduces memory usage by up to 57.8%. Besides, we integrate PMAlloc in a persistent FPTree. Compared to the state-of-the-art allocators, PMAlloc improves the performance of this application by up to 3.1x.

Parallel private information retrieval protocol with index anonymity for untrusted databases

Private information retrieval (PIR) enables a client to search data from a single (or multiple) untrusted server, without revealing which entry was queried. PIR can be divided into two categories: information-theoretic PIR (IT-PIR) and computational PIR (CPIR). However, there is a deployment challenge with IT-PIR because the non-collusion assumption is difficult to implement in practice. Meanwhile, due to the fact that CPIR involves performing cryptographic operations on each element, its performance significantly decreases as the size and number of database entries increase. To overcome these problems, based on homomorphic encryption, this paper presents a parallel PIR (ParPIR) with index anonymity for untrusted databases, which uses the batch encoding to compress request size. Furthermore, our ParPIR eliminates computationally expensive ciphertext multiplication operations, which improves the response time. Moreover, our ParPIR packs multiple responses through rotation column operations on ciphertext, thereby reducing the response size. Compared to previous PIR protocols, the offline time in our ParPIR has almost 1∼63× improvement; the online time in our ParPIR decreases by at least 49.9∼98.9%. Meanwhile, although the request size of our ParPIR has a 0∼31× increase, the response size of our ParPIR has a 0∼4095× improvement.

Optimal and simple approximate solutions to a production-inventory system with two production rates

ABSTRACT We consider a production-inventory system in which the facility produces continuously, switching between two production rates: one faster and one slower than the average demand rate. Demand follows a compound Poisson process, and the size of each demand request is an exponential random variable. Unsatisfied demand is backordered. The production-inventory system is controlled by a two-critical number policy ( r , R ) , whereby production switches from the slower rate to the faster rate when inventory drops below level r and from the faster rate to the slower rate when inventory reaches level R. A fixed cost occurs whenever production switches rates. Our analysis covers two cases: r ≥ 0 and the less studied case of r ≤ 0. We use a level crossing approach to derive the steady-state distribution of the inventory level. Using the steady-state distribution of the inventory level, we calculate the total expected inventory holding, backorder, and switchover costs for each of the two cases. We outline how to obtain the optimal policy through a search of the expected cost functions. We also propose heuristics that give simple closed-form solutions with near-optimal performance. Through a numerical study, we illustrate the importance of considering the r ≤ 0 case.

401 Creating a Dashboard to Increase Efficiency in Tracking and Reporting on Research Data Requests

OBJECTIVES/GOALS: The purpose of the project was to create a Tableau dashboard to track metrics on requests for research data at Atrium Health Wake Forest Baptist. The objectives included: 1) define and identify request fulfillment metrics, 2) build a dashboard to capture metrics, and 3) integrate the dashboard into metrics tracking and reporting activities. METHODS/STUDY POPULATION: Project managers and team leaders in the Office of Informatics collaborated to determine which measures would be most relevant and impactful to report on. Metrics that were collected included: total count of tickets fulfilled over time, number of tickets currently open, sum of outstanding quoted hours, quoted hours vs. actual hours needed to fulfill ticket, and hours billed. Tableau's direct connection feature was used to extract the Trac ticket data from its Postgres database and the dashboard was published to Tableau Server. After the initial draft was created, several rounds of revisions were made as new data insights were discovered through further investigation of the data. RESULTS/ANTICIPATED RESULTS: Each morning, Tableau Server runs an automatic refresh of the data. On the dashboard homepage, users can see a quick view of all available metrics; to minimize noise, only the current statuses, active tickets, and stats for the most recent monitoring periods are displayed. Many of the charts give the user the option to link out to a page with related supplemental information (historic data, ticket status history, etc.). With the help of the dashboard, project managers and team leaders can now monitor how long tickets are in each status, increase quote accuracy using the hours quoted and hours billed charts, and examine ticket complexity over time. DISCUSSION/SIGNIFICANCE: Prior to dashboard creation, metrics were sparse and difficult to assemble. By providing information on the quantity, size, and complexity of data requests, the dashboard enables the Office of Informatics to monitor how the process is functioning overall, make informed decisions about resource allocation, and provide quick interventions.

Perceptions of L1 and L2 Speakers on Task Difficulty of Pragmatic Tasks

In L2 pragmatics, only a few studies have examined task complexity. Furthermore, the existing studies have predominantly focused on the cognitive dimension and have lacked consistent findings. González-Lloret and Ortega (2018) and Pallotti (2019) have thereby contended that socio-interactional features be incorporated into task design. Along this line, this study investigated perceptions of L1 and L2 speakers of English regarding the difficulty of four role-play tasks with differentiated degrees of (dis)preference and imposition. Participants included 33 L1 speakers and 63 Korean L2 speakers at intermediate-level (n = 32) and high-level (n = 31). Results showed that participants’ perceived difficulty matched the design intentions exclusively affected by request size, responsibility for the problem, and persuasion across complex versions of the tasks. Moreover, the linguistic consequences of such factors entailed challenges among L2 speakers. There were also various factors that emerged for task difficulty other than the manipulated task design features, underscoring the importance of participants’ explanatory comments in L2 pragmatics.

Using A Stacked-LSTM Architecture to Analyze API Call Sequences from Dynamic Malware Sandbox

This paper explores the temporal hierarchy of API sequence calls using Deep Recurrent Neural Networks. RNN’s have the capability to easily capture the nature of time series processing due to their sequential structure, whereas generic neural networks cannot find an underlying relationship between data that is separated by timesteps. The primary goal of the paper is to describe an approach to analyzing malware attacks on networks while accounting for each timestep of data and gaining more flexibility in the size of the detection request using the features of the Long Short-Term Memory (LSTM) architecture. We will also touch on the more recent Gated Recurrent Units (GRU) network architecture. We will discuss the tested network configurations and the most optimal model. Lastly, the paper will suggest some applications of this model and how it could be integrated into standard operating systems.

Dynamic Subcarrier Assignment in OFDMA-PONs Based on Deep Reinforcement Learning

Orthogonal Frequency Division Multiplexing Access Passive Optical Networks (OFDMA-PONs), a solution for the next-generation optical access network, allows multiple optical network units (ONUs) to dynamically share subcarriers (SCs) to support efficient bandwidth allocation. In uplink transmission, multiple ONUs can share orthogonal low bit rate SCs to transmit data at different time slots (TSs) during the transmission cycle. In this paper, the dynamic subcarrier allocation (DSA) scheme based on deep reinforcement learning (DRL) is proposed for various ONU bandwidth requests. The novel scheme jointly allocates time slots, subcarriers, and modulation formats in a dynamic and flexible manner. The ONU can save transmit power by using a lower order modulation format while meeting the delay requirement. The simulation part demonstrates how the proposed DRL-based DSA scheme can be adapted to various situations, including 1) variation in the size of ONU bandwidth requests, and 2) variation in the weight of different indicators. The extensive simulation results show that, for the first time, the proposed DRL-based DSA algorithm achieves optimal traffic latency with substantial power saving, compared with the traditional two-dimensional DSA algorithms.

SHA-512 Algorithm on Json Web Token for Restful Web Service-Based Authentication

The development of technology is getting faster and continues to grow so as to create various types of technology, architecture, to new programming languages. Surely this will be a new problem because of differences in technology, programming language, and architecture that must still be able to provide interconnected sources of information. So in order for the system to remain integrated, a Web Service (WS) is needed as a bridge in integrating between systems without differentiating the platform, programming language, or architecture used. One of the Web Service architectures that is widely used is REST (REpresentational State Transfer), but there will be problems in implementing REST Web Service because it does not have security standards in the authentication process. Then an authentication method is needed, namely JSON Web Token (JWT). In implementing JWT, a hash algorithm is needed, such as SHA-512. The results of this study indicate that the use of SHA-512 on the JWT has a good speed with an average data request speed of 512.8 milliseconds (ms) when compared to the SHA-256 algorithm which has an average data request speed of 515.55 MS. Meanwhile, in terms of data size, SHA-512 produces an average data request size of 0.75 kilobytes (kb) compared to SHA-256 which has an average data request size of 0.72 kb.

Shield-Net: Matching Supply with Demand for Face Shields During the COVID-19 Pandemic

The initial months of the COVID-19 pandemic were marked by widespread shortages of personal protective equipment (PPE) because of surging demand and a fragile global supply chain. In response, many domestic suppliers pivoted to producing PPE, such as masks and face shields, made possible by low material costs and simple designs. A key challenge that remained was the lack of an established marketplace for nontraditional suppliers of PPE to connect with healthcare facilities in need. To address this inefficiency, we launched an online platform, Shield-Net, to match requests for face shields with new suppliers of PPE. Our platform was based on an optimization model that produced supplier-requester pairs and took into account request urgency, request size, production capacity, location, and product type. During the period of March to September 2020, Shield-Net produced 390 matches, resulting in the shipment of more than 50,000 face shields to 68 unique requesting organizations. Supplier-requester proximity was found to be the only statistically significant variable in the success of a match. In this paper, we discuss the development and impact of our matching platform, as well as lessons learned during its operation. History: This paper was refereed. This article has been selected for inclusion in the Special Issue on Analytics Remedies to COVID-19. Funding: This work was supported by the National Science Foundation [Grant CMMI-2029072].

A Perfect Knob to Scale Thread Pool on Runtime

Scalability is one of the utmost nonfunctional requirement of server applications, because it maintains an effective performance parallel to the large fluctuating and sometimes unpredictable workload. In order to achieve scalability, thread pool system (TPS) has been used extensively as a middleware service in server applications. The size of thread pool is the most significant factor, that affects the overall performance of servers. Determining the optimal size of thread pool dynamically on runtime is a challenging problem. The most widely used and simple method to tackle this problem is to keep the size of thread pool equal to the request rate, i.e., the frequency-oriented thread pool (FOTP). The FOTPs are the most widely used TPSs in the industry, because of the implementation simplicity, the negligible overhead and the capability to use in any system. However, the frequency-based schemes only focused on one aspect of changes in the load, and that is the fluctuations in request rate. The request rate alone is an imperfect knob to scale thread pool. Thus, this paper presents a workload profiling based FOTP, that focuses on request size (service time of request) besides the request rate as a knob to scale thread pool on runtime, because we argue that the combination of both truly represents the load fluctuation in server-side applications. We evaluated the results of the proposed system against state of the art TPS of Oracle Corporation (by a client-server-based simulator) and concluded that our system outperformed in terms of both; the response times and throughput.

Towards a Blockchain-based Medical Test Results Management System

The role of test results in the diagnosis and treatment of patients’ diseases at medical facilities cannot be ignored. Patients must have a series of tests that are related to their symptoms. This can be repeated as many times as possible, depending on the type of disease and treatment. Seriously, in the cases where the patients lose their medical test record (i.e., patient’s medical history), the diagnosis is difficult due to the lack of information about the medical history as well as the symptoms/complications in the previous treatments. Storing this treatment information in medical centers can address risks related to user failure (e.g., loss of medical test records and wet/fire documents). However, users face a bit of difficulty when they change to other medical centers for medical examination and treatment since the data is stored locally, and difficult to share this with others. Current solutions focus on empowering users (i.e., patients) to share medical information related to disease treatment. However, the main barrier to these approaches is the knowledge of the users. They must embrace some background in terms of the technologies, risks, and rights they may share with treatment facilities. To solve this problem, we propose a Blockchain-based medical test result management system where all information is stored and verified by the stakeholder. The data will be stored decentralized and updated throughout the treatment process. We implement a proof-of-concept based on the Hyperledger Fabric platform. To demonstrate the effectiveness of the proposed system, we conduct evaluation methods based on three main tasks of the system: initializing, accessing, and updating data on six different scenarios (i.e., increasing in size of processing requests). The evaluation based on Hyperledger Caliper helped us to have a deeper analysis of the proposed model.

Evaluating Reliability of SSD-Based I/O Caches in Enterprise Storage Systems

I/O caching techniques are widely employed in enterprise storage systems in order to enhance performance of I/O intensive applications in large-scale data centers. Due to higher performance compared to <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Hard Disk Drives</i> (HDDs) and lower price and non-volatility compared to <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Dynamic Random-Access Memories</i> (DRAM), Flash-based <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Solid-State Drives</i> (SSDs) are used as a main media in the caching layer of storage systems. Although SSDs are known as non-volatile devices but recent studies have reported large number of data failures due to power outage in SSDs. To overcome the reliability implications of SSD-based I/O caching schemes, RAID-1 (mirrored) configuration is commonly used to avoid data loss due to uncommitted write operations. Such configuration, however, may still experience data loss in the cache layer due to correlated failures in SSDs. To our knowledge, <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">none</i> of previous studies have investigated the reliability of SSD-based I/O caching schemes in enterprise storage systems. In this paper, we present a comprehensive analysis investigating the reliability of SSD-based I/O caching architectures used in enterprise storage systems under power failure and high-operating temperature. We explore variety of SSDs from top vendors and investigate the cache reliability in mirrored configuration. To this end, we first develop a physical fault injection and failure detection platform and then investigate the impact of workload dependent parameters on the reliability of I/O cache in the presence of two common failure types in data centers, <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">power outage</i> and <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">high temperature</i> faults. We implement an I/O cache scheme using an open-source I/O cache module in Linux operating system. The experimental results obtained by conducting more than twenty thousand of physical fault injections on the implemented I/O cache with different write policies reveal that the failure rate of the I/O cache is significantly affected by workload dependent parameters. Our results show that unlike workload requests access pattern, the other workload dependent parameters such as request size, <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Working Set Size</i> (WSS), and sequence of the accesses have considerable impact on the I/O cache failure rate. We observe a significant growth in the failure rate in the workloads by decreasing the size of the requests (by more than 14X). Furthermore, we observe that in addition to writes, the read accesses to the I/O cache are subjected to failure in presence of sudden power outage (the failure mainly occurs during promoting data to the cache). In addition, we observe that I/O cache experiences <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">no</i> data failure upon high temperature faults.

Justification for the decision on loading channels of the network of geoecological monitoring of resources of the agroindustrial complex

The article is devoted to solving an urgent scientific and technical problem. It consists of developing an algorithm .This is used for estimating the recommended buffer size of requests of transmitting data streams. So,this has been done in a flying self-organizing network .Therefore, it is used to ensure geoecological monitoring of the resources of the agro-industrial complex. However, Studies have shown that to ensure the required quality of data streams transmission. Thus, it is advisable to carry out preliminary reservation of channel performance. Hence, it is possible to increase the likelihood of servicing requests of the transmission of data streams, as well as to increase the load of network channels by buffering incoming requests of the transmission of data streams. So, the developed algorithm makes it possible to substantiate the decision-making to increase the channel load and ensure an acceptable delay in the start of data streams transmission .Therefore ,it has been done in a flying self-organizing network. Hence, the studies carried out the correctness of the algorithm and the expediency of its application. However, in practice the process of tuning the channels of a flying self-organizing network has been carried.

Justification for the decision on loading channels of the network of geoecological monitoring of resources of the agroindustrial complex

The article is devoted to solving an urgent scientific and technical problem. It consists of developing an algorithm .This is used for estimating the recommended buffer size of requests of transmitting data streams. So,this has been done in a flying self-organizing network .Therefore, it is used to ensure geoecological monitoring of the resources of the agro-industrial complex. However, Studies have shown that to ensure the required quality of data streams transmission. Thus, it is advisable to carry out preliminary reservation of channel performance. Hence, it is possible to increase the likelihood of servicing requests of the transmission of data streams, as well as to increase the load of network channels by buffering incoming requests of the transmission of data streams. So, the developed algorithm makes it possible to substantiate the decision-making to increase the channel load and ensure an acceptable delay in the start of data streams transmission .Therefore ,it has been done in a flying self-organizing network. Hence, the studies carried out the correctness of the algorithm and the expediency of its application. However, in practice the process of tuning the channels of a flying self-organizing network has been carried.

An optimal configuration-based trading scheme for profit optimization in wireless networks

Over the last few years, we have seen dramatic increase in the demand for radio spectrum in the world due to rapid growth in wireless services and applications. In order to meet the growing demand for radio spectrum, new schemes should be developed for utilizing free spectrum effectively. Cognitive radio (CR) is promising technology that enables solving spectrum scarcity problem effectively. Thus, in this paper, we propose two schemes to motivate primary user (PU) to share spectrum with secondary users (SUs). The key concern of the proposed schemes is maximizing PU’s profit by hiring free spectrum to wealthy SUs. In the eviction scheme, SUs with less profit are expelled from service to serve wealthy SUs. The folding scheme reconfigures spectrum request to serve it. It reduces the size of request if the available size is not sufficient for the request. The simulation results demonstrate that the ability of new schemes to maximize PU’s profit. Apparently, the results confirm the ability of the proposed schemes to utilize the free spectrum efficiently.

Data-Level Parallelism Oriented Memory Access and On-Chip Buffering Mechanisms for a Loop Accelerator

Memory access latency is always a bottleneck for the performance improvement of data-intensive applications. Exploiting the memory access patterns of Data-Level Parallelism(DLP) is a promising way for loop accelerators to reduce the latency significantly. This paper proposes two DLP-oriented data provisioning mechanisms to alleviate memory access latency: 1) DLP-oriented memory access(DoMA) for efficiently utilizes the available memory bandwidth. 2) a data access patterns aware on-chip buffer(PABUF) for exploiting reuse in a user-transparent manner. Unlike those loop accelerators using traditional DMA to access global memory, DoMA efficiently reduces the transmission of useless data by adjusting the size of requests intelligently. In addition, PABUF, which manages data using DLP’s memory access patterns without software engineering efforts, allows the loop accelerators to access data in parallel. Experiments show that when our mechanisms are integrated into a loop accelerator based on Rocket Chip Coprocessor(RoCC), it can achieve 4.20x-10.65x(6.81x on average) speedups with negligible overhead of power and area compared to L1 Cache.