In order to achieve high resolution in Successive Approximation Register (SAR) Analog to Digital Converter (ADC), a novel noncoupling capacitor split capacitor array structure is proposed, which solves the coupling capacitor mismatch problem in the traditional split capacitor array structure. This paper analyzes the structure and its capacitor array mismatch. Based on the structure and analysis, a 12-bit SAR ADC is designed and implemented in GSMC 0.11 [Formula: see text]m CMOS process. The measured results show that the Signal to Noise Ratio (SNR) is 72.61 dB and Effective Number Of Bits (ENOB) is 11.77 bit, which prove the advantages of this structure in the design of high resolution SAR ADC.

In order to solve the problems of poor data interaction, low data utilization and weak data security among various systems in the process of smart campus construction, to avoid “information island” and improve data value, this paper carries out an in-depth research from the aspects of data sharing and exchange mode, standard system, data sharing and exchange process. On this basis, this paper puts forward the construction scheme of data sharing and exchange platform. It realizes the data sharing among various business systems, and builds a unified data sharing and exchange center.

Traditional mechanical keys are gradually replaced with touch keys. Capacitive touch keys have advantages in achieving higher sensitivity with a longer service life and lower cost. Hence, the capacitive touch key technology is widely used in consumer electronic products. In use of household electric appliances, such as electric kettle, induction cooker and exhaust hood, sometimes the water vapor would condense into visible moisture or even the water would splash out of the pot when overheated, which may lead to a falsely trigger of touch key. So, waterproof becomes a big challenge in household electric appliances. This paper analyzes how water affects capacitive touch key and proposes a waterproof capacitive touch key sensor interface circuit to overcome the challenge. Sensing of touch key capacitor would be influenced only when water covers more than two touch keys. Compensating channels working in certain strategy are used to decrease the influence of water. For the same water capacitor, the bigger threshold voltage, the bigger change of the counting number. A much smaller threshold voltage is better to further weak the effect of water. The circuit is implemented in a standard 110 nm CMOS process. The measured results show that the touch Signal to Noise and Water Disturb Ration (SNWDR) is 20.7 dB which proves that the proposed waterproof capacitive touch key sensor interface circuit is effective against water splash.

Taking the issue of information island in the field of security surveillance into consideration,this paper proposes a big data-based distributed cloud computing model,consisting of domain node,sub-domain nodes and terminal nodes.The terminal nodes are responsible for data collection,the domain node and sub-domain nodes take charge of data computation and data storage.Then we present an architecture of security surveillance big data based on cloud computing.Characterized by low communication bandwidth,high elasticity of extension,and easy implementation,the proposed computing model integrates the advantages of big data and cloud computing,solves the problem of information island in the field of security surveillance.

针对传统HDFS在存取海量小文件时出现的I/O性能瓶颈问题,本文提出了一种优化的HDFS存取策略.通过文件预处理、增加元数据信息和元数据批处理操作等优化策略,比较好地解决了传统HDFS处理海量小文件的性能问题.并通过试验证明了该方法的有效性.

In the field of big data application, processing the huge amount of data is an issue of great concern and a hard nut to crack in big data application basic research. It is also the main factor that affects the application of big data. Nevertheless, machine learning offers an effective approach to solving this problem. This paper reviews the research on abstract reinforcement learning, factored reinforcement learning, hierarchical reinforcement learning, relational reinforcement learning, and Bayesian reinforcement learning, analyzes their advantages and disadvantages respectively, and points out that combining supervised learning or semi-supervised learning with reinforcement learning is an effective method for machine learning in big data.

针对大数据处理框架MapReduce中的任务调度问题,提出一种基于Markov决策过程(Markov Decision Process,MDP)的任务调度算法,通过状态集来描述集群中节点的负载和作业的数据本地性需求,使用状态转移函数表示调度策略对状态的影响,采用值迭代求解算法求取最优策略,实现集群中节点的最优调度.实验结果表明,该算法能够保证数据本地性的同时,减少作业响应时间,提高系统综合性能.

Focusing on some mistakes in constructing digital campus,presenting the ideology of implementing digital campus is that practical realistic and comprehensive plan.Based on the ideology,Shenzhen Polytechnic Digital Campus,first of all utilized overall planning,including: storage systems,enterprise high-performance servers,database software system,middleware software system,various types of business systems,digital campus portal,unified data center,information security system,operation and maintenance assurance system,information standards,central apparatus room system,and so on.Then pointed out connotation of the digital campus,including: solving island of information,building a unified data center and SOA methodology.Finally a more realistic process control methodology was used to ensure the implementation of the project.

This paper extends the opinion formation theory introduced by Wu and Huberman. Their theory predicts the evolution of opinions through the social network in which individuals are embedded. And they show that the expected weighted fraction of the population which holds a given opinion is constant with respect to time. We assume that each individual A will be affected by one of his neighbors at the probability alpha (A) in the society. In Wu and Huberman's theory, alpha (A) = 1 for any individual A is a special case. However, our assumption is more realistic since not every person will accept other's opinion definitively or at the same probability. Let W-k (B-k) be the set of k-degree individuals which hold the opinion W (B), where the degree k is the sum of the connections from one person to the others. We show that if the distribution of the function a satisfies the condition alpha(k) = kp(k)c for all k, the expected fraction is a martingale, i.e. whose expectation is constant in time. And we believe the constrained condition is too strong and might be weaken. Furthermore, computer experiments are performed to verify our predictions.

This paper is aimed to design and implement cross-platform geometadata engine pool in National Geospatial Data Clearinghouse(NGDC).The geometadata engine pool consists of an integrative geometadata table and a series of APIs/applications/services,with which data users can check the integrative metadata table.With Extensible Markup Language(XML),a text-based markup language which stores data in a structured format by using meaningful tags,developers can further describe geometadata by defining meaningful tags for geometadata applications.Document Object Model(DOM),which is used to access and manipulate the integrative geometadata table based on XML,defines the logical structure of documents and the way a document is accessed and manipulated,and provides an Application Programming Interface(API) for XML documents.