Guo-Heng Luo

A parallel Bees Algorithm implementation on GPU

Bees Algorithm is a population-based method that is a computational bound algorithm whose inspired by the natural behavior of honey bees to finds a near-optimal solution for the search problem. Recently, many parallel swarm based algorithms have been developed for running on GPU (Graphic Processing Unit). Since nowadays developing a parallel Bee Algorithm running on the GPU becomes very important. In this paper, we extend the Bees Algorithm (CUBA (i.e. CUDA based Bees Algorithm)) in order to be run on the CUDA (Compute Unified Device Architecture). CUBA (CUDA based Bees Algorithm). We evaluate the performance of CUBA by conducting some experiments based on numerous famous optimization problems. Results show that CUBA significantly outperforms standard Bees Algorithm in numerous different optimization problems.

Physical Access Control Based on QR Code

Nowadays, the authentication methods of physical access control such as key, smart card, and OTP (One Time Password) card require extra hardware, and it is not easy to be aware of if they are lost because of the low using rate. Nowadays, mobile phone is one of the most common used and carry-on device. It would be detected in a shorter time when a mobile phone is lost because it was frequently used in daily time. One of the advantage of OTP is that it is only used once and it is securer compared to the fixed password. QR code is a kind of two-dimension matrix code, which has high encryption and decryption speed; it can not only store large information but also has the error-correction ability. In this paper, we try to develop a safe and efficient authentication way by using mobile phone and implement it for access control system. We take the advantage of one-time usage by using OTP, and store the secret information into QR code. Therefore, the proposed system enhances the security of physical access control systems.

A decision tree using CUDA GPUs

Classification is an important issue both in Machine Learning and Data Mining. Decision tree is one of the famous classification models. In the reality case, the dimension of data is high and the data size is huge. Building a decision in large data base cost much time in computation. It is a computationally expensive problem. GPU is a special design processor of graphic. The highly parallel features of graphic processing made todays GPU architecture. GPGPU means use GPU to solve non-graphic problems which need amounts of computation power. Since the high performance and capacity/price ratio, many researches use GPU to process lots computation. Compute Unified Device Architecture (CUDA) is a GPGPU solution provided by NVIDIA. This paper provides a new parallel decision tree algorithm base on CUDA. The algorithm parallel computes building phase of decision tree.

Vehicle Driving Video Sharing and Search Framework Based on GPS Data

The driving dispute is a critical problem for drivers when the car accident is happened. The drivers usually install the car video recorder in their car to recode their driving images for many years ago. If a car accident is happened, the driver can provide a driving video file as an evidence to claim that they did not do any dangerous driving, and protect themself. However, in some situations, drivers may not in the car accident when driving the car, but they have the same requirement for a video record, because they want to use those video files to find out the crime in hit-and-run accident. Due to an evolution of social network, many people were post the required messages to find the driving videos recorded in some specific times and locations. In fact, such kind of messages can be beneficial for many people to solve the hit-and-run accident by using social networks and driving videos. The goal of this paper is to develop a framework which can provide a platform for users to upload their driving videos, and allow other users can search video by a given specific time, date and location from the frameworks’ database. This framework can also provide the application on mobile devices for user to recorder their driving videos, and this application can upload their driving video into the frameworks’ database automatically.

Auto-Scaling Mechanism for Cloud Resource Management Based on Client-Side Turnaround Time

Currently, providers of Software as a service (SaaS) can use Infrastructure as a Service (IaaS) to obtain the resources required for serving customers. SaaS providers can save substantially on costs by using resource-management techniques such as auto scaling. However, in most current auto-scaling methods, server-side system information is used for adjusting the amount of resources, which does not allow the overall service performance to be evaluated. In this paper, a novel auto-scaling mechanism is proposed for ensuring the stability of service performance from the client-side of view. In the proposed mechanism, turnaround time monitors are deployed as clients outside the service, and the information collected is used for driving a dynamic auto-scaling operation. A system is also designed to support the proposed auto scaling mechanism. The results of experiments show that using this mechanism, stable service quality can be ensured and, moreover, that a certain amount of quality variation can be handled in order to allow the stability of the service performance to be increased.

Personalized Cloud Storage System: A Combination of LDAP Distributed File System

“Cloud computing” gradually flourish, a wide range of distributed storage systems are increasingly diverse, Like of Gluster, Ceph, Lustre, as well as Hadoop, etc.. In this paper, we propose a personal cloud storage system Integrated with pNFS, it can be accessed in parallel for scalable performance. Besides, data backup and failover mechanism are designed. We expect that the function of the proposed system, not only can improve performance, but also can improve the user’s privacy in order to provide better service quality in cloud storage.

File Deduplication with Cloud Storage File System

The Hadoop Distributed File System (HDFS) is used to solve the storage problem of huge data, but does not provide a handling mechanism of duplicate files. In this study, the middle layer file system in the HBASE virtual architecture is used to do File Deduplicate in HDFS, with two architectures proposed according to different requires of the applied requirement reliability, therein one is RFD-HDFS (Reliable File Deduplicated HDFS) which is not permitted to have any errors and the other is FD-HDFS (File Deduplicated HDFS) which can tolerate very few errors. In addition to the advantage of the space complexity, the marginal benefits from it are explored. Assuming a popular video is uploaded to HDFS by one million users, through the Hadoop replication, they are divided into three million files to store, that is a practice wasting disk space very much and only by the cloud to remove repeats for effectively loading. By that, only three file spaces are taken up, namely the 100% utility of removing duplicate files reaches. The experimental architecture is a cloud based documentation system, like the version of EndNote Cloud, to simulate the cluster effect of massive database when the researcher synchronized the data with cloud storage.

Cloud Resource Management With Turnaround Time Driven Auto-Scaling

Cloud resource management research and techniques have received relevant attention in the last years. In particular, recently numerous studies have focused on determining the relationship between server-side system information and performance experience for reducing resource wastage. However, the genuine experiences of clients cannot be readily understood only by using the collected server-side information. In this paper, a cloud resource management framework with two novel turnaround time driven auto-scaling mechanisms is proposed for ensuring the stability of service performance. In the first mechanism, turnaround time monitors are deployed in the client-side instead of the more traditional server-side, and the information collected outside the server is used for driving a dynamic auto-scaling operation. In the second mechanism, a schedule-based auto scaling preconfiguration maker is designed to test and identify the amount of resources required in the cloud. The reported experimental results demonstrate that using our original framework for cloud resource management, stable service quality can be ensured and, moreover, a certain amount of quality variation can be handled in order to allow the stability of the service performance to be increased.

A P2P Auto-Recover Method for the Factory Data of Cable Modem over the HFC Network

With the popularization of the Data-Over-Cable Service Interface Specifications (DOCSIS) product, many devices have been installed in client sides. the issue of how to maintain the quality while providing quick repair becomes more and more important. in some practical circumstances, the cable modem may lose the factory data which include the calibration table and the digital certificates. These data are unique and vital for the regular operations of cable modem. the traditional way of handling the malfunction product is to ship back the product to the maintenance branch and replaces a working device for the customer. in this paper we will present a solution which is based on network coding over peer to peer (P2P) network to fix the problem automatically online, named Cable modem Auto-Recovery (CAROL).

A Two-Layer SVM Classification Mechanism for Chinese Blog Article

In Taiwan, the famous bloggers can be regard as professional writers now. More and more people subscribe their RSS (Really Simple Syndication) to receive updated information. But readers might only interest in few categories of articles, readers need to filter other articles by themselves. In order to help people select the information they want, this research proposed a two-layer SVM classification mechanism to classify blog articles. The schema is also evaluated in this research and the experiment result the proposed schema achieves 87% of recall and 95% of precision.