Search

In distributed storage systems, erasure codes represent an attractive data redundancy solution which can provide the same reliability as replication requiring much less storage space. Multiple data losses happens usually and the lost data should be regenerated to maintain data redundancy in distributed storage systems. Regeneration for multiple data losses is expected to be finished as soon as possible, because the regeneration time can influence the data reliability and availability of distributed storage systems. However, multiple data losses is usually regenerated by regenerating single data loss one by one, which brings high entire regeneration time and severely reduces the data reliability and availability of distributed storage systems. In this paper, we propose a tree-structured parallel regeneration scheme based on regenerating codes (TPRORC) for multiple data losses in distributed storage systems. In our scheme, multiple regeneration trees based on regenerating code are constructed. Firstly, these trees are created independently, each of which dose not share any edges from the others and is responsible for one data loss; secondly, every regeneration tree based on regenerating codes owns the least network traffic and bandwidth optimized-paths for regenerating its data loss. Thus it can perform parallel regeneration for multiple data losses by using multiple optimized topology trees, in which network bandwidth is utilized efficiently and entire regeneration is overlapped. Our simulation results show that the tree-structured parallel regeneration scheme reduces the regeneration time significantly, compared to other regular regeneration schemes.

Spectrum trading is an important approach for the secondary users (SU, the unlicensed user) to share the spectrum resource with the primary users (PU, the licensed user). Such economic method can improve not only the efficiency but also the quality of spectrum sharing. For getting the most profit from the spectrum trading, we considered a spectrum trading model based on queueing theory. We analyzed the relationship between the spectrum server and the spectrum demander and discussed the problem of server uncertainty. Simulation results demonstrated the analysis results in terms of the idle probability and average queue size.

The ever-growing demand on information and data requires the efficient architecture for large-scale network storage systems. To serve very large scale applications, using inexpensive commodity becomes the common selection in nowadays cloud storage systems. Based on such unreliable hardware, building fault-tolerant mechanism is key issue to the system design. In this paper, we propose a rack-aware architecture for cloud storage systems.

In order to fully utilize the spectrum resource, dynamic spectrum access becomes a promising approach to increase the opportunity of spectrum access with the rapid development of cognitive radio. However the performance of cognitive radio networks (CRN) is considerably constrained by its limited power, memory and computing ability. Fortunately cloud computing, which is the highlight of current research, has the potential to make up for the disadvantages because of its vast storage and computing capacity. In this paper we will discuss the convergence of spectrum sharing and cloud computing from several aspects including model, advantages and challenges. A spectrum sharing model based on cloud computing (SSC) will be introduced.

In wireless cognitive networks the secondary users (SU, the unlicensed user) can share the spectrum via opportunistic access and spectrum trade. The former can’t offer the reliable and steady communication, while the latter can. It is a hot topic how the primary users (PU, the licensed user) provide the resources to SUs. We propose a novel spectrum trade model based on the Voronoi diagram: V-STM, which can promise fairness and rationality of spectrum lease. We also analyze the effect of communication range and price on spectrum trade. Simulation results demonstrate the performance of the proposed model in terms of the fairness of users and the whole profit of the network.

Various data-centric web applications are becoming the developing trend of information society. Cloud computing currently adopt column-oriented storage wide table to represent the heterogeneous structured data of these applications. The wide table reduces the waste of storage space, but slows down query efficiency. The paper implements the hybrid partition on access frequent (HPAF) to horizontally and vertically partition a wide table. It uses a variant of consistent hashing to dynamically horizontally partition a wide table across multiple storage nodes on each node’s performance; It use entropy to represent the number of reducing access data block from the table with N columns than from N column-oriented storage tables. According to the second law of thermodynamics, the paper designs an entropy increasing clustering algorithm to classify the columns of a wide table. The algorithm finds a cluster with multiple classes which save maximum access time. The paper implements an algorithm for structured query across multiple materialized views too. Lastly the paper demonstrates the query performance and storage efficiency of our strategy compared to single column storage.

In this study, Ni-P-SiC composite coatings were prepared by adding different amount of SiC particles into electroless Ni-P plating solution. The effects of SiC particles concentration in solution on depositing rate and composition of the coatings were investigated. The different SiC contents in coatings could be obtained by adjusting the amount of SiC particle concentration in electroless plating solution. Scanning electron microscopy (SEM) results indicate that SiC particles are dispersed in the composite coating and Ni-P alloy is mainly as a bond metal envelope with SiC particles. At the same time, the microhardness became increased and the wear resistance improved with the increasing of SiC content.

Data distribution is the basic behavior of P2P applications ( le sharing and stream- ing service) and it is a key element a ecting the performance of P2P systems. However, there are few research works that focus on data distribution of P2P applications from the view of whole system. In this paper we study the data distribution in P2P applications in terms of de- creasing the system distribution load. We de ne the distribution load of P2P systems formally and analyze how to decrease the system load quickly by means of mathematical analysis. More- over, we give a feasible fast distribution algorithm according to our theoretic conclusion. The experimental results show that our algorithm has signi cant improvement on data distribution speed and load balance.

A wear tester was developed. MC PA (nylon) filled with MoS2 and PU (polyurethane) were used as the material of the rubber wheel of roller guide shoes. Their wear performances was investigated with the tester. The results show that the wear rate of MC PA increases firstly and then decreases with increasing load, and reverses with increasing velocity. The wear rate of PU decreases firstly and then increases with increasing velocity. In addition, the main wear mechanisms of PU are plough wear and abrasive wear at the low load (200N). At the loads of 200N~500N, the main wear form of MC PA is the adhesive wear. Adhesive wear is the main mechanism of MC PA at the low velocity (3m/s). Due to inner heat accumulation by friction, squama-peering occurs on MC PA surface at the high velocity (8m/s). The dominant wear mechanisms of PU are abrasive wear and fatigue pitting. And the main reason of PU’s failure is the interior heat accumulation caused by friction.