Redundant Storage Research Articles

Анализ избыточности хранения темпоральных данных средствами реляционных СУБД

Анализ избыточности хранения темпоральных данных средствами реляционных СУБД

Exploiting Symmetry in Tensors for High Performance: Multiplication with Symmetric Tensors

Symmetric tensor operations arise in a wide variety of computations. However, the benefits of exploiting symmetry in order to reduce storage and computation is in conflict with a desire to simplify memory access patterns. In this paper, we propose a blocked data structure (Blocked Compact Symmetric Storage) wherein we consider the tensor by blocks and store only the unique blocks of a symmetric tensor. We propose an algorithm-by-blocks, already shown of benefit for matrix computations, that exploits this storage format by utilizing a series of temporary tensors to avoid redundant computation. Further, partial symmetry within temporaries is exploited to further avoid redundant storage and redundant computation. A detailed analysis shows that, relative to storing and computing with tensors without taking advantage of symmetry and partial symmetry, storage requirements are reduced by a factor of $ O\left( m! \right)$ and computational requirements by a factor of $O\left( (m+1)!/2^m \right)$, where $ m $ is the order of the tensor. However, as the analysis shows, care must be taken in choosing the correct block size to ensure these storage and computational benefits are achieved (particularly for low-order tensors). An implementation demonstrates that storage is greatly reduced and the complexity introduced by storing and computing with tensors by blocks is manageable. Preliminary results demonstrate that computational time is also reduced. The paper concludes with a discussion of how insights in this paper point to opportunities for generalizing recent advances in the domain of linear algebra libraries to the field of multi-linear computation.

Technology Scaling Comparison of Flip-Flop Heavy-Ion Single-Event Upset Cross Sections

Heavy-ion experimental results from flip-flops in 180-nm to 28-nm bulk technologies are used to quantify single-event upset trends. The results show that as technologies scale, D flip-flop single-event upset cross sections decrease while redundant storage node flip-flops cross sections may stay the same or increase depending on the layout spacing of storage nodes. As technology feature sizes become smaller, D flip-flop single-event upset cross sections approach redundant storage node hardened flip-flops cross sections for particles with high linear energy transfer values. Experimental results show that redundant storage node designs provide > 100X improvement in single-event upset cross section over DFF for ion linear energy transfer values below 10 MeV-cm2/mg down to 28-nm feature sizes.

A New Organization and Indexing Method of Multibeam Point Cloud Data in 3D Marine GIS

Multibeam sounding system can be used for the acquisition of high-precision digital seabed terrain model, it brings new opportunities for the development of 3D marine GIS (Geographic Information System). But its massive point cloud data poses great challenge for its storage and visualization. This paper proposes a new storage strategy and gives an optimized visualization method specifically for the storage strategy. First, we take advantage of the point cloud coordinates regularity to sort these points according to certain rules and efficiently organize it through multiple spatial index. Vertically, a multi-level pyramid index is established. We generate a number of levels data with different level of detail and different scale, namely the static level of detail (LOD) model, in this way, a trade-off between appropriate redundant storage space and efficient data retrieve is achieved. Horizontally, for all levels of the pyramid model, we establish a tiled quadtree index, and generate a cache image for each tile. Secondly, an optimized visualization method in accordance with the storage strategy is also designed. Based on the current area of interest and data dispatch algorithm, we calculate the corresponding tile location and directly load the cache image to lessen the workload of complex spatial data operation, reduce the rendering burden of spatial data. We can generate customized dynamic cache with specific area of interest and scale, to improve the visual effect, too.

A dual redundancy radiation-hardened flip—flop based on a C-element in a 65 nm process

A radiation-hardened flip—flop is proposed to mitigate the single event upset (SEU) effect. Immunity was achieved through the use of C-elements and redundant storage elements. It takes advantage of the property of a C-element in which it enters a high impedance mode when its inputs are of different logic values. Redundant storage nodes are then used to drive the C-elements so that a single upset pulse in any storage will be prevented from altering the state of the output of the flip—flop. The flip—flop was implemented using 48 transistors and occupied an area of 30.78 μm2, using 65 nm CMOS process. It consumed 22.6% fewer transistors as compared to the traditional SEU resilient TMR flip—flop.

8K UHD(7680×4320) H.264/AVC 부호화기를 위한 4×4블럭단위 보간 필터 및 SAD트리 기반 부화소 움직임 추정 엔진 설계

본 연구에서는 8K UHD(7680×4320) 영상을 실시간 부호화하기 위한 4×4 블록 부화소 움직임추정기를 제안한다. 연산처리성능을 향상시키기 위해 보간 연산을 4×4 블록 단위로 병렬화시켰으며, 병렬 보간 연산에서 필요한 메모리 대역폭을 확장하기 위해 10×10개의 메모리 어레이를 가진 2D 캐쉬 버퍼 구조를 설계하였다. 그리고 2D 캐쉬 버퍼는 검색영역 간 재사용 기법을 적용하여 참조화소의 중복저장을 최소화하였으며, 4×4 블록 병렬 보간 필터는 3단(수평·수직 1/2부화소, 대각선 1/2부화소, 1/4 부화소) 평면 보간 연산 파이프라인 구조로 설계하여 연산회로를 고속화시켰다. 0.13㎛ 공정에서 시뮬레이션한 결과, 436.5K게이트의 4×4 블록 부화소 움직임추정기는 동작주파수 187㎒에서 8K UHD급 동영상을 초당 30프레임으로 실시간 처리하는 성능을 보였다.

4K UHD급 H.264/AVC 복호화기를 위한 4×4 블록 병렬 보간 움직임보상기 아키텍처 설계

In this paper, we proposed a block parallel architecture of interpolation for high-performance H.264/AVC Motion Compensation in 4K UHD() video real time processing. To improve throughput, we design block parallel interpolation. For supplying the reference data for interpolation, we design 2D cache buffer which consists of the memory arrays. We minimize redundant storage of the reference pixel by applying the Search Area Stripe Reuse scheme(SASR), and implement high-speed plane interpolator with 3-stage pipeline(Horizontal Vertical 1/2 interpolation, Diagonal 1/2 interpolation, 1/4 interpolation). The proposed architecture was simulated in 0.13um standard cell library. The maximum operation frequency is 150MHz. The gate count is 161Kgates. The proposed H.264/AVC Motion Compensation can support 4K UHD at 72 frames per second by running at 150MHz.

Three Design Principles of Language: The Search for Parsimony in Redundancy

In this paper we present three design principles of language - experience, heterogeneity and redundancy--and present recent developments in a family of models incorporating them, namely Data-Oriented Parsing/Unsupervised Data-Oriented Parsing. Although the idea of some form of redundant storage has become part and parcel of parsing technologies and usage-based linguistic approaches alike, the question how much of it is cognitively realistic and/or computationally optimally efficient is an open one. We argue that a segmentation-based approach (Bayesian Model Merging) combined with an all-subtrees approach reduces the number of rules needed to achieve an optimal performance, thus making the parser more efficient. At the same time, starting from unsegmented wholes comes closer to the acquisitional situation of a language learner, and thus adds to the cognitive plausibility of the model.

Outage at UAA

With redundant hardware, it is rare that a disk failure results in downtime at the system level. System failures do sometimes occur, typically as a sequence of very rare events that leads to a catastrophic failure. This case describes how a combination of hardware and firmware failures, along with human error, led to the failure of a redundant disk storage unit, which in turn affected several enterprise systems at a major public university. Subsequently, a small number of conservative and seemingly “good” decisions in the process of restoring the system from backups led to negative outcomes, primarily additional downtime over the course of several days. The case illustrates how even well-considered and conservative decisions may seem flawed in hindsight. An important lesson from the case is that it is difficult to justify to management the provision of sufficient backup resources to prevent very low-probability failure events.

Intelligent Decision Making Technique for Marketing Using Hypothetical Database and Fuzzy Multi-Criteria Method

Marketing organizations use databases to locate potential customers and to generate sales lead. A number of software systems have been playing a key role in supporting the decision making activities in recent years but common problem in all that they cannot handle fuzzy data appropriately for "what-if" analysis. This paper proposes intelligent scenarios analysis system framework for marketing decision support which deals with crisp and fuzzy data like linguistic variable. Applying new approach "Hypothetical Database" for derived data that permits decision manager to manage views according to the need of organization and/or market environment. View in Hypothetical Database provides versatility in "What-If" analysis by using versions of "What-If" database and reduce data redundancy and data storage in updating. Using Fuzzy database will help to handle imprecise and uncertain information like "Linguistic variable" in a more human oriented process. Finally, the projected scenarios selected by decision manager will be aligned in a hierarchy according to the distance from "Ideal Vector" by using Fuzzy Multi-Criteria Decision Making method.

The CaMKII/NMDAR complex as a molecular memory

CaMKII is a major synaptic protein that is activated during the induction of long-term potentiation (LTP) by the Ca2+ influx through NMDARs. This activation is required for LTP induction, but the role of the kinase in the maintenance of LTP is less clear. Elucidating the mechanisms of maintenance may provide insights into the molecular processes that underlie the stability of stored memories. In this brief review, we will outline the criteria for evaluating an LTP maintenance mechanism. The specific hypothesis evaluated is that LTP is maintained by the complex of activated CaMKII with the NMDAR. The evidence in support of this hypothesis is substantial, but further experiments are required, notably to determine the time course and persistence of complex after LTP induction. Additional work is also required to elucidate how the CaMKII/NMDAR complex produces the structural growth of the synapse that underlies late LTP. It has been proposed by Frey and Morris that late LTP involves the setting of a molecular tag during LTP induction, which subsequently allows the activated synapse to capture the proteins responsible for late LTP. However, the molecular processes by which this leads to the structural growth that underlies late LTP are completely unclear. Based on known binding reactions, we suggest the first molecularly specific version of tag/capture hypothesis: that the CaMKII/NMDAR complex, once formed, serves as a tag, which then leads to a binding cascade involving densin, delta-catenin, and N-cadherin (some of which are newly synthesized). Delta-catenin binds AMPA-binding protein (ABP), leading to the LTP-induced increase in AMPA channel content. The addition of postsynaptic N-cadherin, and the complementary increase on the presynaptic side, leads to a trans-synaptically coordinated increase in synapse size (and more release sites). It is suggested that synaptic strength is stored stably through the combined actions of the CaMKII/NMDAR complex and N-cadherin dimers. These N-cadherin pairs have redundant storage that could provide informational stability in a manner analogous to the base-pairing in DNA.

Compression of Multi-Resolution Terrain Data Based on Binary Tree

In order to overcome the incontinuity and redundancy of data storage caused by the grids in representing the multi-resolution terrain, an adaptive hierarchical triangulation based on binary tree is used to represent the multi-resolution terrain model. It realizes the representation of the terrain data at different levels of detail and enables data compression by local omission of data values. The structure of binary tree is stored using a bit code of the underlying binary tree, while the height data are stored using an array, and relative pointers which allow a selective tree traversal. This method realizes the continuity and reduces the data volume in data storage of multi-resolution digital elevation model (DEM), and it is possible to work directly on the compressed data. We show that significant compression rates can be obtained already for small threshold values, and in a visualization application, it is possible to extracted and drawn triangulations at interactive rates.

Video Sequence Matching Based on the Invariance of Color Correlation

Video sequence matching aims to locate a query video clip in a video database. It plays an important role in reducing storage redundancy and detecting video copies for copyright protection. In this paper, we propose an effective method for video sequence matching based on the invariance of color correlation. The proposed method first splits each key-frame into nonoverlapping blocks. For each block, we sort the red, green, and blue color components according to their average intensities, and use the percentage of the color correlation to generate a frame feature with a small size. Finally, the resulting video feature is made up of the consecutive frame features, which is demonstrated to be robust against most typical video content-preserving operations, including geometric distortion, blurring, noise contamination, contrast enhancement, and strong re-encoding. The experimental results show that the proposed method outperforms the existing methods in the literature, as well as the method based on the traditional color histogram. Furthermore, the time and space complexity of our algorithm are both satisfactory, which are very important for many real-time applications.

Hierarchical RAID: Design, performance, reliability, and recovery

Hierarchical RAID (HRAID) extends the RAID paradigm to mask the failure of whole Storage Nodes (SNs) or bricks, where each SN is a disk array with a certain RAID level. HRAIDk/ℓ with N SNs and M disks per SN tolerates k SN failures and ℓ disk failures per SN with Maximum Distance Separable (MDS) erasure codes, which introduce the minimum level of redundancy at each level. For N=M there are k internode and ℓ intranode check strips per SN, occupying the capacity of as many disks with storage redundancy (k+ℓ)/N, but a higher storage redundancy is required for M>N. HRAIDk/ℓ tolerates all disk failures up to dmin=(k+1)(ℓ+1)−1, but up to dmax=Nℓ+Mk−kℓ disk failures can be tolerated. Three options for HRAID operation are: (I) Only intranode recovery. (II) Intranode and internode recovery on demand reconstruction of blocks and rebuild. (III) Multistep internode recovery with no rebuild processing. The I/Os Per Second (IOPS) metric is used to assess the cost of fault-tolerance for HRAIDk/ℓ against RAID(4+ℓ) and RAID0, for varying k and ℓ. The maximum IOPS is at its lowest in degraded mode, but even with fewer operational disks the normal mode IOPS may be exceeded after restriping. Asymptotic reliability analysis and simulation results show that HRAIDk/ℓ with ℓ>k provides a higher reliability when SN failures are due to disk rather than controller failures. Monte Carlo simulation is used to quantify the effect of various recovery options with varying k and ℓ and as the SN controller failure rate is varied with respect to disk failure rates on the Mean Time to Data Loss (MTTDL). The HRAID paradigm is justified by the fact that Options II attains a significantly higher MTTDL than Option I. Option III with no rebuild processing has an MTTDL exceeding Option II, but a poorer performance.44This paper combines and significantly extends the following works [35,41,42].

Liénard-type dynamical system models for the simulation of the human heart atrium

on the instantaneous health status of the patient. Most of these data are lost soon after it is recorded. The objective of this project is to develop a computing cluster to record and process all physiological data from all ICU bedside monitors within a hospital. Methods: Custom informatics server software was written to link 10 off-the-shelf commodity computers into a cluster computing system. An SQL database was constructed to organize the physiological and clinical data of patients. A graphical user interface was constructed in MATLAB to allow for easy data inspection, visualization, and markup. Physiological data are distributed to the nodes of this system each night. Results: The informatics cluster computing system has been online since January 2009 at the University of Virginia Medical Center. The system contains 80 processors, 40 GB of RAM and 100 TB of hard-drive capacity, including redundant data storage. Approximately 100 MB of data are recorded per bed per day and are composed of 3 electrocardiograms, chest impedance, oxygen saturation, heart rate, blood pressure, as well as other vital signs and waveforms. The system contains a largely complete physiological record of the approximately 1200 patients admitted to the neonatal ICU, pediatric ICU, and surgical trauma ICU since the system became operational. This system is currently being rolled out at Texas Children's Hospital. Conclusions: We have successfully deployed a cluster computing system to capture and process physiological data from bedside monitors within an ICU environment. We are using this system to develop new metrics for patient health, which can eventually be presented to the physician at the bedside to better guide their care.

Building an Open Data Repository for a Specialized Research Community: Process, Challenges and Lessons

In 2009, the Institution for Social and Policy Studies (ISPS) at Yale University began building an open access digital collection of social science experimental data, metadata, and associated files produced by ISPS researchers. The digital repository was created to support the replication of research findings and to enable further data analysis and instruction. Content is submitted to a rigorous process of quality assessment and normalization, including transformation of statistical code into R, an open source statistical software. Other requirements included: (a) that the repository be integrated with the current database of publications and projects publicly available on the ISPS website; (b) that it offered open access to datasets, documentation, and statistical software program files; (c) that it utilized persistent linking services and redundant storage provided within the Yale Digital Commons infrastructure; and (d) that it operated in accordance with the prevailing standards of the digital preservation community. In partnership with Yale’s Office of Digital Assets and Infrastructure (ODAI), the ISPS Data Archive was launched in the fall of 2010. We describe the process of creating the repository, discuss prospects for similar projects in the future, and explain how this specialized repository fits into the larger digital landscape at Yale.

NORMALIZATION IN TYPE-2 FUZZY RELATIONAL DATA MODEL BASED ON FUZZY FUNCTIONAL DEPENDENCY USING FUZZY FUNCTIONS

In fuzzy relational databases, the data dependencies, especially the fuzzy functional dependency(ffd) plays an important role in maintaining the consistency of the database and in avoiding the redundant storage of the data. In the past, it has been shown that the type-2 fuzzy relational databases captures impreciseness and incompleteness in data in a better way. The aim of this paper is to provide the concepts for database normalization in a type-2 fuzzy relational database, so that the normalized schemas can be obtained. Here, we deal with the fuzzy functional dependency(ffd) based normalization of type-2 fuzzy relational databases. We use the concepts of fuzzy functions to derive the fuzzy equality and using this fuzzy equality, we define a new definition of fuzzy functional dependency. First we discuss various approaches proposed by the researchers in this context and show why our fuzzy functional dependency is better, as compared to the earlier ffds proposed by the researchers. We call our ffd as non-0 LHS ffd. We identify an anomaly called "spurious ffd" and show that some of the significant contributions proposed by the earlier researchers are suffering from this anomaly, but the non-0 LHS ffd does not suffer from it. Then, we prove that the set of inference rules for the non-0 LHS ffd are sound and complete. We use the definition of non-0 LHS ffd in obtaining the first three normal forms upto BCNF for type-1 and type-2 fuzzy relational schemas. The result of the decomposition and the procedure to obtain the membership value of the decomposed relations is proposed. The associated concepts like the fuzzy key, fuzzy superkey, fuzzy foreign key are defined in terms of non-0 LHS ffd. On the basis of these concepts, we define full ffd, partial ffd etc. In the last, we show that in our case, the relationship of total-ordering between the three normal forms in classical relational databases is also observed.

Spatial Access Method for Flash Memory Embedded Systems

This paper proposed an efficient spatial Access method, called MDR-Tree (Mbr compression and Delayed write operation based R-Tree), that uses the node compression technique and the delayed write operation technique for flash memory embedded systems. The node compression technique of MDR-Tree increased the utilization of flash memory space by compressing the MBR of spatial data using relative coordinates and MBR size. Moreover, the delayed write operation technique reduced the number of write operations in flash memory by temporarily storing spatial data in the buffer and by reflecting them in flash memory at once instead of reflecting the insert, update and delete of spatial data in flash memory for each operation. Especially, the utilization of buffer space was enhanced by preventing the redundant storage of the same spatial data in the buffer.

An Approach for Optimizing Relational Provenance Storage

Modern data management has to deal with data from different sources with different quality, therefore, supporting data provenance in the system level and allowing users to know where data comes from and how it was derived have become a critical research topic. Annotation is one of approaches to track provenance. However, storing fine-grained annotations can be expensive as the complete annotations for the data may outsize the storage space required for the data itself. In this paper, we propose a framework for storing provenance information relating to data derived via relational queries, using provenance trees which match the query structure to avoid redundant storage of information about the derivation process. Within this framework, we come up with a series of storage optimization methods against the relational queries to make good choices of query tree nodes where provenance information should be stored. Our optimization algorithms run in time polynomial in the query size and linear in the size of the provenance, thus enabling provenance tracking and optimization without incurring large overheads. This framework is a new idea for the data tracing study and has a wide range of applications.

Chord-Based Indexing Model to Support Complex Query and Load Balancing

This paper presents a new model (SMF-Chord) which is expanded from Chord. SMF-Chord uses double-fingerprint, double-mapping and dynamic forwarding mechanism to support multi-attribute multi-keyword fuzzy-matching query, and also has a load balancing mechanism which includes three parts: similar-close transposition, forward balancing and hot-set cache. The experiment results show SMF-Chord has high recall rate with low storage redundancy, and it can effectively balance load when node mapping load, file storage load, or query load is unbalanced.