In a computational Grid which consists of many computer clusters, job start time predictions are useful to guide resource selections and balance the workload distribution. However, basic middleware available today either has no means expressing that site will take before starting or uses simple linear scale. this paper we introduce system for predicting times on clusters. Our based statistical analysis historical traces simulation schedulers. We have deployed EDG (European Data-Grid)...

As an important partner of fifth generation (5G) communication, the internet things (IoT) is widely used in many fields with its characteristics massive terminals, intelligent processing, and remote control. In this paper, we analyze security performance for cooperative non-orthogonal multiple access (NOMA) networks IoT, where multi-relay Wyner model direct link between base station eavesdropper considered. particular, secrecy outage probability (SOP) two kinds relay selection (RS) schemes...

In large-scale grids with many possible resources (clusters of computing elements) to run applications, it is useful that the can provide predictions job response times so users or resource brokers make better scheduling decisions. Two metrics need be estimated for time predictions: one how long a executes on (application time), other waits in queue before starting (queue wait time). this paper we propose an instance based learning technique predict these two by mining historical workloads....

In this paper we present an initial analysis of job failures in a large-scale data-intensive Grid. Based on three representative periods production, characterize the interarrival times and life spans failed jobs. Different failure types are distinguished is carried out further at Virtual Organization (VO) level. The spatial behavior, namely where occur Grid, also examined. Cross-correlation structures, including how arrivals correlate with failures, analyzed illustrated. We investigate...

Local learning has been proposed as a common framework to predict both application run times and queue wait based on workload traces [8]. The time is shown be more difficult expensive because its distance calculations typically involve not only job attributes but also resource states. In this paper methods algorithms are investigated improve prediction accuracy performance for times. Firstly, the so-called "local tuning" adopted tune parameters each training subset divided by pivot attribute...

Deep learning algorithms offer an effective solution to the inefficiencies and poor results of traditional methods for building a footprint extraction from high-resolution remote sensing imagery. However, heterogeneous shapes sizes buildings render local vulnerable influence intricate backgrounds or scenes, culminating in intra-class inconsistency inaccurate segmentation outcomes. Moreover, extracting very (VHR) images at present often lose spatial texture information during down-sampling,...

In this paper we present an initial analysis of job failures in a large-scale data-intensive Grid. Based on three representative periods production, characterize the interarrival times and life spans failed jobs. Different failure types are distinguished is carried out further at Virtual Organization (VO) level. The spatial behavior, namely where occur Grid, also examined. Cross-correlation structures, including how arrivals correlate with failures, analyzed illustrated. We investigate...

This paper analyzed the development of data mining and fifth generation (5G) for Internet Things (IoT) uses a deep learning method stock forecasting. In order to solve problems such as low accuracy training complexity caused by complicated in model forecasting, we proposed forecasting based on feature selection (FS) Long Short-Term Memory (LSTM) algorithm predict closing price stock. Considering its future potential application, this takes 4 from Shenzhen Component Index an example...

An improved projection algorithm was put forward in the process of segmentation license plate characters. According to feature that trough created by space between second and third character very stable, detected as starting point for locating left right boundary width characters fixed, when one side located, a value jumped firstly then other located traditional algorithm. The experimental result shows compared algorithm, is not sensitive noise things at sides plate. And it can correctly...

A novel CDMA based multiple-access scheme for differential chaos shift keying (DCSK) is proposed in this paper. By using the orthogonal pseudo-random codes, multi-user access ability can be provided both synchronized and asynchronous DCSK system. Firstly, transmitter receiver structure of CDMA-DCSK system are given, which with slight modification to original system, basic idea principle also described. Then bit error rate performance analyzed, several formulae evaluating BER obtained....

In order to overcome shortcomings of basic hidden Markov model (HMM), a hybrid multi-layer perceptron (MLP) and continuous (CHMM) is presented which bases on HMM. this mode, MLP calculates each state's output probability instead CHMM. The main purpose improve the recognition ratio CHMM by means strong MLP's nonlinear predictive capability. Speaker independent Mandarin digit speech based models realized. Experimental results show that efficiency has higher than

To solve the problem of considerable deformation tailgate in a fully mechanized caving face, position main roof fracture line is first obtained by theoretical calculation, combined with results similar simulation test and numerical analysis. The width section coal pillar determined to be 11.5 m. Based on distribution characteristics deviatoric stress plastic zone surrounding rock, location geological conditions, new zoning control design approach proposed, “two pillars, three zones, parts”...

Key-evolving cryptography is intended to mitigate the damage in case of a secret key compromise, one severest security threats actual cryptographic schemes. In public-key setting, essential idea key-evolving lies updating private with time, while maintaining same public key. Key evolution encryption and signing has been well studied, especially identity-based (ID-based) setting where an entity's can be derived from that identity information. From more practical standpoint, however, would...

In this paper, a novel facial expression analysis system is proposed, which directly recognizes warp image of face. And the can be obtained by automatic AAM processing. The consists mainly four modules, processing, Gabor filtering, Adaboost training SVM and testing We also compared performance our method to trained on output face detector (geometry normalization images). using preprocessing reach high level accuracy above 90%