Search

International currency market weathers all forms of worldwide current affairs and governmental economics statistics releases. Such news causes fluctuations in demands and supplies for major currencies. In order to gain from currency tradings, a trader has to gauge, sometimes guess using gut feelings, the likely market movement, and act accordingly. In this paper, a neural network-based expert decision support system for assisting users in making currency option trading decisions is presented. By utilizing neural network technology in its inference engine, the system is able to learn new knowledge and refine its inference strategy through usage. Furthermore, it can change its reasoning strategies according to different users, thus realizing the idea of personalized logic system.

This research introduces predictive models that are based on N-tier software development architecture exploiting the new technologies. The desire to predict software readiness is essential to any software house. DOCTOR OF PHILOSOPHY (EEE)

The availability of high performance smartphones had generated much interest in services that integrates computing capabilities into daily life. Research in this area focuses on making consumer devices context aware and appropriate to the circumstances. This thesis describes the development of an application distribution manager for non-persistent mobile application. We looked at different aspects of localization technologies available today and how these techniques are implemented in positioning systems. There is a gap in research on mobile service application distribution using location as the context. The thesis considers context modelling as a tool for improving ubiquitous computing. Context modelling is an important aspect in localization as it dictates the flow of information input into a system and the corresponding information outflow. Such a modelling tool can help develop a structured approach to handle location information. We describe a context refinement scheme that helps to manage context information while supporting user autonomy and control. We list the practical consideration on how the context refinement scheme can be applied to mobile localization based on the environment in which it operates. We looked at challenges associated with mobile environment and the proposals to mitigate such problems. Signal drag is a key issue identified to be resolved for the implementation of a proximity based mobile distribution service. A marked improvement in reducing signal drag in a wireless medium is observed using a proposed cluster based approach. The proposed cluster based solution is more efficient in identifying the relative contributions from wireless data. It is incorporated into the localization aspect of the proposed proximity distribution mechanism. The system is capable of distributing relevant service applications when in proximity to a smart space The prototype proximity based software distribution platform tags applications to smart service areas. Smart spaces are predefined areas that provide related service applications in the vicinity. This application distribution framework achieved context awareness by pushing only relevant applications to users when they are in a service area. The software prototype was tested in a real world setting using windows based and android based platforms and the results showed good recognition rate for areas of interest based on user’s context and interest. MASTER OF ENGINEERING (EEE)

Bibliometrics, citation analysis, journal impact fact or and Hirsch’s index are techniques and metrics used to study communication pattern and evaluate research performance of researchers in the academic world. This study uses those techniques and metrics, together with the aid of Journal Citation Reports (JCR) 2006 and Scopus 2006 citation database, on 2963 researchers from business, computer science and electrical engineering disciplines from 2 local universities in Singapore and 8 Ivy League universities in the United States. We have also expanded this study to cover additional 274 researchers from computer science discipline from 3 leading Asian universities in Hong Kong and Japan and 3 leading Commonwealth universities in Australia and the United Kingdom. This study establishes the 2 facts that computer science researchers communicate more using journal papers, rat her than the conference papers, and that local universities in Singapore are lagging behind Ivy League universities in term of research performance. The expanded study establishes 2 additional facts that local universities in Singapore are comparable to Asian universities in term of research performance in computer science discipline, and that local universities in Singapore are ahead of Commonwealth universities in tern of research performance in computer science discipline. The expanded study also reaffirms the fact that computer science researchers communicate more using journal papers, rather than conference papers. Master of Science (Communication Software and Networks)

The Federal Reserve determines federal funds target rate (FFTR), which is one of the most publicized and anticipated economic indicator in the financial world. As the decision making process is complex due to unknown functions, it has been a difficult and challenging process for the researcher to model the thoughts of the FOMC members using statistical methods and hence predict the changes in FFTR. With artificial neural networks evolving as an efficient and promising methodology, it is possible to emulate the decision making of FOMC. Master of Science (Communication Software and Networks)

83 p. Mobile phones have come to be seen as a convergence device. They are being sought after by the rich and the not so rich alike. Mobile devices are constant companions during our everyday routines. Information and services useful to a user varies with respect to his/her immediate 'situation'. Situation here refers to context. A context-aware service is capable of delivering contextually relevant informatioa By doing so, we are inching towards making the ubiquitous mobile phones more personal, resourceful and convenient. Master of Science (Communication Software and Networks)

98 p. Online banking and e-commerce experienced rapid growth over the past few years and show tremendous promise of growth even in the future. This has made it easier for fraudsters to indulge in new and abstruse ways of committing credit card fraud over the electronic channels. Statistics of the losses incurred due to credit card fraud show that technology is yet to catch up with the new trends of fraud and every new step towards effective fraud detection is of immense commercial value to the whole business-to- consumer industry. Master of Science (Communication Software and Networks)

69 p. Stock Selection using GGAP-RBF (General Growing and Pruning Radial Basis Function) Neural Network In the stock picking or selection problem, the motivation is to design system such that it is able to systematically select stocks or equities that are empirically predicted to perform well in the future. Thus, it should achieve four objectives: 1) to achieve monetary returns in the future; 2) to beat the average market performance; 3) to avoid the bias in human-based stock selection due to psychological reasons; 4) automatically crunching the data to extract useful information based on past information, which can be hardly done manually. Hence, this dissertation presents methodologies to select equities based on soft-computing models and the information gathered from fundamental analysis of stock performance. Master of Science (Communication Software and Networks)

Today, there are various electronic payment systems out there for all sort of electronic commerce transactions. Electronic transaction ranges from few cents is refer as Micro payments. Master of Science (Communication Software and Networks)

The focus of this work is to study the feasibility and practicality of using neural networks as a forecasting tool so as to determine if detectable trends are present in stock performance, particularly on initial public offers. Finance and investing is the second most frequent business area of neural networks applications after productiodoperations. Although many research results show that neural networks can solve almost all problems more efficiently than traditional modeling and statistical methods, there are opposite research results showing that statistical methods in particular data samples outperform neural networks. Many papers on neural network applications on stock markets provide forecast only on existing stocks. However, many new stocks are being listed each year. Thus the aim of this study is to explore this relatively un-tapped region in the stock market and to check if neural networks can make the prediction of returns of these IPOs. Master of Science (Communication Software and Networks)

This project involves the development of an E-learning master-proxy system for the hosting and retrieval of E-learning documents. eavily accessed documents are dynamically distributed via mobile agents to proxy hosts surrounding the main E-learning server. Some of the heavy traffic can then be re-routed away from the central server, easing the bottleneck. The faded information field layer structure method was studied and verified. The response time of this method is observed, analysed and compared to traditional method. It was found that there is significant improvement in response time. This dissertation concludes by suggesting additional experimental works and tests for refining the system. Master of Science (Communication Software and Networks)

In this thesis we proposed and implemented a mobile agent assisted e-learning framework, in which mobile agent based searching and information filtering have been added to the existing e-learning system to realize faster information retrieval. Master of Engineering

Study how market-based model uses prices to coordinate resource allocation for mobile agents and hosts, based on local information and interests and the framework and requirements of secured mobile agent systems, and the security issues and infrastructures required to adopt the pricing techniques securely. Master of Science (Communication Software and Networks)

This research starts from the survey and study of software measurement and software metrics, nineteen metrics which belong to seven categories are identified. Then an effective, affordable metrics extraction tool for C++ programs, METC++, has been developed. The set of identified metrics can be extracted by this tool. Master of Science (Communication Software and Networks)

This thesis proposes an approach for rule-based personalization system. Through analyzing the based characters of rule-based personalization systems, this project decomposes the rules and provides a kind of atomic components named rule item. Users can use rule items to make up of rules for their own personalized profile. On the other hand, the system also traces the "mouse trails" to silently follow the visitors of the web site and finds the relationships between the actions of visitors. Based on these relationships, the system will automatically generate the rules for each visitor using the rule items according to his habit. Master of Science (Communication Software and Networks)

The research project proposes methods for an adaptive WAP portal which will help to reduce consumers' time spent in navigation and with the time saved for fruitful content browsing. The methods had also taken into account possible drift of consumers' interest over time and used weighted computation to achieve adaptation that will be of relevance to consumers at any point of time. Master of Science (Communication Software and Networks)

The background of this project is an e-learning system using mobile agent(s) for information gathering. The e-learning system has made a great achievement in information gathering and retrieval. Master of Science (Communication Software and Networks)

In this project, the student will survey the existing push technology that has been written in literatures, reexamining current push technology in commercial activity, identify the problems facing and investigate the possibility of proposing new solution to solve the limitation of current push technology. Master of Science (Communication Software and Networks)

one size fits one Master of Science (Communication and Network Systems)

This dissertation provides a literature survey of electronic payment systems that can be grouped into five broad classes: credit-card based systems, electronic cheque, centralized account based systems, electronic cash and micropayment systems. Master of Science (Communication Software and Networks)

This project proposes and implements a system for seamless integration of HP E-speak E-services and IBM Aglet mobile agent applications onto a common platform. And it conducts a systematic comparative study between E-speak, Aglets, related technologies, and E-Commerce requirements. Master of Science (Communication Software and Networks)

Several factors determine the mechanisms deployable for congestion control in ATM networks. The Leaky Bucket is recognized as one of the best forms of traffic regulators for controlling the rate at which the source generates traffic. When a source enters into a traffic contract with the network, it specifies different traffic characteristics like peak cell rate, sustainable cell rate, minimum cell rate and maximum burst size. As long as the source submits traffic within the limits specified in the contract, the policing mechanism remains inactive. When the contract is abused, the policing mechanism comes into effect. The violating cells are dropped or tagged and thus, preventing the network from getting congested further downstream. In this research, the effect of Leaky Bucket policing on Real-Time Variable Bit Rate traffic and Non-Real-Time Variable Bit Rate traffic is explored. The Cell Loss Ratio, ATM Virtual Circuit Load and TCP Traffic delay are also investigated.* Master of Science (Communication and Network Systems)

The capability of neural networks to control connection admission in Asynchronous Transfer Mode (ATM) networks is investigated. The general problem of connection admission control (CAC) and its formulation as a functional mapping are discussed, leading to applications of neural networks and their associated algorithms to the solution of CAC problems. In particular, the use of the class of feed-forward neural networks with backpropagation learning rule and the learning vector quantization (LVQ) network are being studied. Various frameworks have been proposed for the ATM traffic control, but it is not easy to build an efficient traffic control system because of the diversity in multimedia traffic characteristics. This diversity complicates the traffic control system, and various assumptions and simplified traffic models are required to design a practical system based on the traditional mathematical calculations and computer simulations. Neural networks are thought to have many potential applications in ATM traffic control. In this research, the major aim is to present and to compare different neural architectures applicable to connection admission control* Master of Engineering

This research describes an application of Case-Based Reasoning (CBR) for admission control in Asynchronous Transfer Mode (ATM) network and what is required to make CBR works. In this research, one of the Case-Based Reasoning Tool called CASe Parser by IAN (CASPIAN) is used. In this tool, it uses CASpian Language (CASL) as a language for building the case file which is executed by this Case-Based Reasoning Tool. The case pool is used to bottle up past experiences in dealing with similar problems during the connection set-up where the admission control in the ATM network function is to decide whether the new connection can be accepted or not based on whether the Quality of Service (QoS) will be violated as a result. In this research, cell loss probability is used as a gauge of the Quality of Service (QoS). Master of Science (Communication and Network Systems)

The goal of using traffic shaping is to maximize the usage or throughput of ATM networks especially in the area of multimedia. Traffic shaping is part of the set of procedures in ATM traffic management whose primary objective is to protect the ATM network and end-systems from congestion in order to promote efficient use of network resources. Master of Science (Communication and Network Systems)

The mechanisms deployable for congestion control in ATM networks depend on a variety of factors including the arrival pattern of cells and the QoS guarantees required. It can be inferred that the preventive or open loop congestion control scheme is likely to be more effective due to the large ratio of propagation time to transmission time prevalent in high-speed networks. Master of Science (Communication and Network Systems)

Broadband Integrated Service Digital Networks (B-ISDN), along with Asynchronous Transfer Mode (ATM), are designed to support the requirements of modern communication that cover multiple users with multimedia traffic including voice, data, video. The resource allocation by the ATM Connection Admission Control (CAC) is one of the crucial factors needed to meet the certain criteria of the network performance that must guarantee the quality of service negotiated of the existing connection. To develop a reliable ATM Connection Admission Controller is a very difficult task, because of the dynamic nature of the source traffic. The associated mathematical algorithm is complex. Master of Science (Communication and Network Systems)

Traffic shaping is used in ATM networks to achieve a desired modification of traffic characteristics. It is part of the set of procedures in ATM traffic management whose primary role is to protect the ATM network and end-systems from congestion in order to achieve network perform objectives. An additional role of traffic management is to promote the efficient use of network resources. Master of Science (Communication and Network Systems)