1,720,979 research outputs found
The link level security implementation of bluetooth technology /
No full textBluetooth technology is a de facto global standard for a short-range, low-power and low-cost form of wireless communication. New as it is, the technology is anticipated to play a very important role in an increasingly mobile world.Like almost any other modern communication technology, Bluetooth must be sufficiently secure before it can be widely accepted. This is particularly challenging for Bluetooth due to the fact that it uses radio as the physical communication media. The nature of microwave links, as opposed to fixed cable connections, means that Bluetooth is especially vulnerable to attacks. However, through the use of some well-known and highly-effective cryptographic algorithms, Bluetooth is able to achieve the three most important security goals: secrecy, authenticity and integrity.The focus of the thesis is how Bluetooth translates the theoretical cryptographic techniques into effective real life implementations. Security measures in Bluetooth are taken at both the application level and the link level. In the thesis, the author places the emphasis on the link level measures. Various security areas are discussed, such as random number generation, key management, encryption, authentication, etc. The limitations and weaknesses of Bluetooth security implementations are also discussed, and some suggestions for future improvements are offered
Parallel algorithms for visibility analysis and path planning
No full textIn search and rescue (SAR) and emergency planning operations real-time response is critical, particularly when lives are endangered. Until recently, such operations were planned manually by experts in the field using tools such as templates, protractors and calculators. An initiative from the Department of National Defence of Canada has recommended the development of a prototype computerized tool to help the experts in the planning phase of these operations.This thesis, which was carried out under PRECARN project Passport, addresses two types of algorithms that are at the core of SAR and emergency planning operations: airborne path planning for surveillance and search, and land path planning for mobility analysis and rescue route determination.The major limitation associated with these algorithms is the high computational cost incurred from the treatment of vast amounts of data representing elevation, transportation networks, vegetation and other soil characteristics. In addition, the programming complexity of these algorithms is not negligible. These issues demand the high performance computing capabilities and relative ease of programming of today's multiprocessor shared-memory computers, as well as algorithms that are scalable and highly efficient.Parallel algorithms for visibility analysis and path planning using digital terrain data are proposed. The algorithms have been implemented on a Sun symmetric multiprocessor (SMP) SPARC 20 machine and integrated with the Passport airborne path planning and land path planning software. Results of experiments are presented. Some results are compared with a previous implementation on the MasPar MP-1 and MP-2 massively parallel computers
Detection of faulty components in object-oriented systems using design metrics and a machine learning algorithm
No full textObject-Oriented (OO) technology claims faster development and higher quality of software than the procedural paradigm. The quality of the product is the single most important reason that determines its acceptance and success. The basic project management problem is "delivery of a product with targeted quality, within the budget, and on schedule". We propose a state-of-the-art approach that gets closer to the solution by improving the software development process used. An important objective in all software development is to ensure that the delivered product is as fault-free as possible. We proposed three hypotheses that relate the OO design properties---inheritance, cohesion, and coupling---and the fault-proneness as software's quality indicator. We built classification models that predict which components are likely to be faulty, based on an appropriate suite of OO design measures. The models represent empirical evidence that the aforementioned relationships exist. We used the C4.5 machine learning algorithm as a predictive modeling technique, because it is robust, reliable, and allows intelligible interpretation of the results. We defined three new measures that quantify the specific contribution of each of the metrics selected by the model(s), and also provide a deeper insight into the design structure of the product. We evaluated the quality of the predictive models using an objective set of standards. The models built have high quality
Performance analysis of computer systems
No full textWith an ever-growing and more productive computer industry, the performance of computer systems has become a major concern. Problems related to computer performance usually occur either because the system is not correctly sized or because its resources are not adequately allocated.Performance analysis of computer systems is the process of evaluating the current performance of a system by monitoring and studying its behavior under different loads. It involves a deep understanding of the functioning of the basic components of a system. Performance analysis is typically followed by performance tuning, in which required changes are applied to the system in order to achieve optimum performance.In this thesis, we discuss the basics of performance analysis. The different resources of a system are described and an overview of performance-monitoring tools for these resources is presented. An application of performance analysis is also included: two new major systems at McGill University are analyzed (the library management system ALEPH and the finance system BANNER)
Variables abstraction in SDL
No full textState explosion problem is an essential impediment in practical applications of formal languages such as SDL, Specification and Description Language. In the context of validation and testing, one of the possible solutions is to abstract a constructed formal specification by removing certain, unimportant from some viewpoint, variables. The resulting system could be more tractable by existing validation and testing tools. In this thesis, we first elaborate methods for abstracting variables in the Extended FSM model and then methods to treat SDL specifications that are based on the EFSM model. The methods are implemented in a tool that is successfully tested on a real life SDL specification
Distance education on the world-wide web
No full textThe Internet's growing acceptance as a key medium to educate people is the main focus of this thesis. Although distance education has been a popular learning method since the end of the 19th Century, never before has technology made it so easy to disseminate knowledge by linking different media formats---e.g. text, sound, video. In fact, this huge library accessed by a universal interface, is one of the key contributions that the Internet and the World-Wide Web have brought to the learning process.The main goal of this thesis is to encourage instructors to create and deliver courses using the Internet and, most of all, to show that the process can be simple and effective. In order to support this study, four major Course Management tools are presented and analyzed: Pathway by Solis-Macromedia, LearningSpace by Lotus, WebCT by The University of British Columbia, and Virtual-U by Simon Fraser University.As a result of this thesis, a Grades Application was developed using the Internet protocol. This application is an uncomplicated, yet effective solution for using the Web to manage, calculate, and view students' marks. With the open architecture of the Web and standard programming languages such as JavaScript and Perl, the system will execute in most computers available in universities around the world
Paperless assignments : a closer look
No full textWith the increasing student population at many tertiary level institutions, the management of assignment submissions and results has become a challenging feat. This thesis proposes that one way to address these challenges is to develop methods for automating the submission of assignments to support the administrative effort. This will not only enable many administrative tasks to be automated, but will also support and in some cases, improve the learning experience available to students.The features of four paperless assignment submission systems are presented, compared and analyzed with the goal of exposing instructors to the varied benefits that such systems can provide, thereby encouraging them to use such systems in their course delivery.It is also the intent that this thesis will form the basis from which further developmental work will be performed on such systems in general, and on VisualCM in particular. (VisualCM is a paperless assignment submission system currently used within the School of Computer Science at McGill University)
A policy based network configuration framework /
No full textNetworks and their services have been growing rapidly in recent years, so has the complexity of configuring the network operations. In order to manage the rapidly expanding networks, new management protocols are thus needed to change the network management from configuration of individual devices to automation of the whole network. SNMPCONF, a policy-based configuration with SNMP protocol, is proposed by IETF to meet this requirement.SNMPCONF is suggested to configure networks with a policy MIB (Management Information Base). However, deploying this protocol in existing network devices needs more work, because the policy MIB defined by SNMPCONF is not implemented now in the present network and no configuration MIB defined. Even if the configuration MIB and policy MIB implemented by vendors in future, some existing devices may not be upgraded to support these MIB for their hardware limitation (e.g., memory limit). Another problem is found in the communication between a SNMP supported network and the SNMP unsupported networks (e.g. telecommunication network where TL1 is used in North America). In this study, CLI and TL1 are used to help solving these problems. Based on Modular SNMP from University of Quebec at Montreal, a JAVA framework to deploy SNMPCONF with this approach is implemented and tested on a network built with Cisco routers (Cisco IOS 12.0, routers ranged from Catalyst 2600 to 2900). The preliminary work shows the SNMPCONF can be implemented by wrapping CLI commands as accessory functions in policy language without configuration MIB. As shown in this work, another advantage of this approach is that policy execution is atomic and persistent
Financial trading systems - neural and genetic algorithms
No full textIn today's financial markets, when new information is disseminated with lightning speed across the investment community, individual investors turn to trading systems as a way to generate profit. Based primarily on Technical Analysis, a trading system can take advantage of a plethora of advanced modeling tools available today ranging from chart pattern recognition to genetic optimization of technical indicators and trading rules. This paper offers a systematic approach to financial system development involving neural networks and genetic algorithms. A trading system that forecasts S&P500 index is developed and analyzed
Temporal-oriented policy-driven network management
No full textSystems management represents the set of activities necessary to ensure that information systems function according to user requirements and objectives. Chapter 1 summarizes the management challenges in today's networks and distributed systems. Policy-driven network management is the new management paradigm. Its implementation needs a new information and decision model, appropriate protocols and new hosting and access mechanisms. IETF policy framework and architecture create the support for the deployment of this new paradigm.Consideration of temporal aspects to allow run-time policy conflict detection and error-handling has not yet been developed. After assessing the existing policy-oriented achievements in terms of tools and IETF drafts, and after presenting existing temporal mechanisms, we concluded that only policy definition temporal issues are partially referred to. We considered complementary temporal issues focusing on "policy execution" and coping with the notions of time zones, policy dependency across multiple time zones and actions translation among many time zones. Finally, we showed how our input can be used for extending the current IETF Policy CIM proposal.We intend to have an IETF draft on these issues. Specifically, our proposal can be added as a new building block to any existing policy-enabled management tool. We identified important directions in handling policy conflicts at run-time
- …
