Considerando-se os design methods ou os métodos de projetação arquitetônica disponíveis, a presente dissertação aborda as gramáticas da forma (shape grammars) como sistema de simulação, geração e otimização de formas. Essas gramáticas podem combinar recursos descritivos e prescritivos para a solução de problemas de projeto, principalmente por meio da geração de tipos, padrões e famílias de projeto, destinados aos contextos do design industrial e da produção arquitetônica pré-fabricada. Particularmente, o manejo de gramáticas de projeto podem qualificar projetos, cujos programas envolvam uma reprodução em escala, em que se faz necessário criar variedade, flexibilidade e versatilidade construtiva. Este objeto de estudo compõe o arcabouço teórico essencial para uma reflexão a respeito de implementações computacionais e para a automação do projeto arquitetônico. Utilizando a base teórica e a reflexão resultante, esta pesquisa procura um rebatimento prático dos fundamentos teórico-conceituais implícitos, por meio de uma análise da linguagem e da gramática de projeto presentes na arquitetura Industrializada do arquiteto João Filgueiras Lima, o Lelé, focalizando suas primeiras escolas pré-fabricadas, construidas no período de 1983 a 1993. Espera-se, com isso contribuir para que arquitetos e usuários criem e compartilhem repertórios de projeto e de processos criativos, por meio de plataformas computacionais mais amigáveis à projetação arquitetônica, as quais tomem como base, sobretudo, as especificidades desse campo do conhecimento. / Considering the design methods and the architectural design methods available, this dissertation approaches shape grammars as a system of forms simulation, generation and optimization. These grammars can combine descriptive and prescriptive resources for solving project problems, mainly through the generation of project types, standards and families, in the contexts of industrial design and prefabricated architectural production. Particularly, the handling of project grammars can qualify projects whose programs involve a reproduction in scale, where variety, flexibility and constructive versatility are necessary. This object of study makes up the essential theoretical framework for a reflection on computational implementations and for the automation of an architectural project. Using the theoretical basis and the resulting reflection, this research aims at stimulating a practical debate on the implicit theoretical and conceptual foundations, through an analysis of the language and the project grammar present in architect João Filgueiras Lima (Lele)´s industrialized architecture, focusing on his first prefabricated schools, built during the period 1983-1993. We expect to offer some contribution to architects and users so they can create and share project repertoires and creative processes through friendlier computing platforms to architectural design, which should be based, mainly, on the particularities of this knowledge field.

Considerando-se os design methods ou os métodos de projetação arquitetônica disponíveis, a presente dissertação aborda as gramáticas da forma (shape grammars) como sistema de simulação, geração e otimização de formas. Essas gramáticas podem combinar recursos descritivos e prescritivos para a solução de problemas de projeto, principalmente por meio da geração de tipos, padrões e famílias de projeto, destinados aos contextos do design industrial e da produção arquitetônica pré-fabricada. Particularmente, o manejo de gramáticas de projeto podem qualificar projetos, cujos programas envolvam uma reprodução em escala, em que se faz necessário criar variedade, flexibilidade e versatilidade construtiva. Este objeto de estudo compõe o arcabouço teórico essencial para uma reflexão a respeito de implementações computacionais e para a automação do projeto arquitetônico. Utilizando a base teórica e a reflexão resultante, esta pesquisa procura um rebatimento prático dos fundamentos teórico-conceituais implícitos, por meio de uma análise da linguagem e da gramática de projeto presentes na arquitetura Industrializada do arquiteto João Filgueiras Lima, o Lelé, focalizando suas primeiras escolas pré-fabricadas, construidas no período de 1983 a 1993. Espera-se, com isso contribuir para que arquitetos e usuários criem e compartilhem repertórios de projeto e de processos criativos, por meio de plataformas computacionais mais amigáveis à projetação arquitetônica, as quais tomem como base, sobretudo, as especificidades desse campo do conhecimento. / Considering the design methods and the architectural design methods available, this dissertation approaches shape grammars as a system of forms simulation, generation and optimization. These grammars can combine descriptive and prescriptive resources for solving project problems, mainly through the generation of project types, standards and families, in the contexts of industrial design and prefabricated architectural production. Particularly, the handling of project grammars can qualify projects whose programs involve a reproduction in scale, where variety, flexibility and constructive versatility are necessary. This object of study makes up the essential theoretical framework for a reflection on computational implementations and for the automation of an architectural project. Using the theoretical basis and the resulting reflection, this research aims at stimulating a practical debate on the implicit theoretical and conceptual foundations, through an analysis of the language and the project grammar present in architect João Filgueiras Lima (Lele)´s industrialized architecture, focusing on his first prefabricated schools, built during the period 1983-1993. We expect to offer some contribution to architects and users so they can create and share project repertoires and creative processes through friendlier computing platforms to architectural design, which should be based, mainly, on the particularities of this knowledge field.

In this thesis, we consider how service platforms can provide personalized service to incoming consumers in order to improve ease and convenience for their users, or to enhance their own revenues. With an increasing trend toward digitalization, there is now a massive amount of data that can be leveraged to accomplish this goal. In this thesis, we explore how to leverage feature data about the consumers and products, as well as the way consumers interact with the platforms, in order to make better operational decisions, such as pricing, ranking, and recommendations of products and services. In Chapters 1 & 2, we study platforms in which consumers' purchasing decisions are strongly influenced by reviews that are posted by previous consumers. In Chapter 1, we consider how a platform can learn from the choices that consumers make, as well as the reviews they leave, in order to form increasingly accurate estimates about product quality and consumer preferences, so that the platform can provide better personalized product rankings. In Chapter 2, we focus on platforms where each consumer forms an impression of a product by browsing the available reviews in a chosen order following a cascade click model. We consider how to rank the reviews as well as how to price a product given the reviews, in order to maximize short- and long-term revenue. In Chapter 3, we study a platform that offers a menu of memberships of different durations to a pool of heterogeneous consumers. A consumer's choice depends on both the menu prices and his personal intended usage time. The platform wants to maximize the long-term revenue while considering the choices made by heterogeneous customers. We show insights into the optimal solution of each problem when all the parameters are known. For the case where some parameters are initially unknown, we propose a joint learning and optimization algorithm, and provide theoretical guarantees for its performance.

A framework for the quantitative analysis of performance, security and database management within a network system (e.g. a cloud computing platform) is presented within this research. Our study provides a methodology for modelling and quantitatively generating significant metrics needed in the evaluation of a network system. To narrow this research, a study is carried-out into the quantitative modelling and analysis of performance, security and database management trade-offs in cloud computing platforms, based on Stochastic Activity Networks (SANs) and combined metrics. Cloud computing is an innovative distributed computing archetypal based on the infrastructure of the internet providing computational power, application, storage and infrastructure services. Security mechanisms such as: batch rekeying, intrusion detection, encryption/decryption or security protocols come at the expense of performance and computing resources consumption. Furthermore, database management processing also has an adverse effect on performance especially in the presence of big data. Stochastic Activity Networks (SANs) that offer synchronisation, timeliness and parallelism are proposed for the modelling and quantitative evaluations of ‘optimal’ trade-offs involving performance, security and database management. Performance modelling and analysis of computer network systems has mostly been considered of utmost importance. Quantification of performance for a while has been assessed using stochastic models with a rising interest in the quantification of security stochastic modelling being applied to security problems. Quantitative techniques that includes analytical valuations founded on queuing theory, discrete-event simulations and correlated approximations have been utilised in the examination of performance. Security suffers from the point that no interpretations can be made in an optimal case. The most consequential security metrics are in analogy with reliability metrics. The express rate at which data grows increases the prominence for research into the design and development of cloud computing models that manages the workload intensity and are suitable for data exploration. Handling big data especially within cloud computing is a resource consuming, time-demanding and challenging task that necessitates titanic computational infrastructures to endorse successful data exploration. We present an improved Security State Transition Diagram (SSTD) by adding a new security state (Failed/Freeze state). The presence of this new security state signifies a security position of the computing network system were the implemented security countermeasures cannot handle the security attacks and the system fails completely. In a more sophisticated security system, when the security countermeasure(s) cannot in any form categorise the security attack, the network system is moved to the Failed/Freeze security state. At this security state, the network system can only resume operation when restored by the system administrator. In this study, we propose a cloud computing system model, defined security countermeasures and evaluated the optimisation problems for the trade-offs between performance, security and database management using SANs formalism. We designed, modelled and implemented dependency within our presented security system, developing interaction within the security countermeasures using our proposed Security Group Communication System (SGCS). The choice of Petri-Nets enables the understanding and capturing of specified metrics at different stages of the proposed cloud computing model. In this thesis, an overview of cloud computing including its classification and services is presented in conjunction with a review of existing works of literature. Subsequently, a methodology is proposed for the quantitative analysis of our proposed cloud computing model of performance-security-database trade-offs using Möbius simulator. Additionally, numerical experiments with relevant interpretations are presented and appropriate interpretations are made. We identified that there are system parameters that can be used to optimise the presented abstract combined metrics but they are optimal for neither performance or security or database management independently. Founded on the proposed quantitative simulation model framework, reliable numerical experiments were observed and indicated scope for further extensions of this work. For example, the use of Machine Learning (ML) or Artificial Intelligence (AI) in the predictive and prevention aspects of the security systems.

A challenge posed by future computer architectures is the efficient exploitation of their many and sometimes heterogeneous computational cores. This challenge is exacerbated by the multiple facilities for data movement and sharing across cores resident on such platforms. To answer the question of how systems software should treat heterogeneous resources, this dissertation describes an approach that (1) creates a common manageable pool for all the resources present in the platform, and then (2) provides virtual machines (VMs) with multiple `personalities', flexibly mapped to and efficiently run on the heterogeneous underlying hardware. A VM's personality is its execution context on the different types of available processing resources usable by the VM. We provide mechanisms for making such platforms manageable and evaluate coordinated scheduling policies for mapping different VM personalities on heterogeneous hardware. Towards that end, this dissertation contributes technologies that include (1) restructuring hypervisor and system functions to create high performance environments that enable flexibility of execution and data sharing, (2) scheduling and other resource management infrastructure for supporting diverse application needs and heterogeneous platform characteristics, and (3) hypervisor level policies to permit efficient and coordinated resource usage and sharing. Experimental evaluations on multiple heterogeneous platforms, like one comprised of x86-based cores with attached NVIDIA accelerators and others with asymmetric elements on chip, demonstrate the utility of the approach and its ability to efficiently host diverse applications and resource management methods.

Thesis (MScEng)--Stellenbosch University, 2013. / ENGLISH ABSTRACT: The way in which people consume video is changing with the adoption of new technologies such as tablet computers and smart televisions. These new technologies, along with the Internet, are moving video distribution away from satellite and terrestrial broadcast to distribution over the Internet. Services online now offer the same content that originally was only available on satellite broadcast television. However, these services are only viable in countries with high speed, inexpensive Internet bandwidth. The need therefore exists for alternative services to deliver content in countries where bandwidth is still expensive and slow. These include many of the developing nations of Africa. In this thesis we design and develop a video distribution platform that relies on peer-to-peer networking to deliver high quality video content. We use an existing video streaming peer-to-peer protocol as the primary distribution mechanism, but allow users to share video over other protocols and services. These can include BitTorrent, DC++ and users sharing hard drives with one another. In order to protect the video content, we design and implement a security scheme that prevents users from pirating video content, while allowing easy distribution of video data. The core of the security scheme requires a low bandwidth Internet connection to a server that streams keys to unlock the video content. The project also includes the development of a custom video player application to integrate with the security scheme. The platform is not limited to, but is aimed at high speed local area networks where bandwidth is free. In order for the platform to support feasible business models, we provision additional services, such as video cataloging and search, video usage monitoring and platform administration. The thesis includes a literature study on techniques and solutions to secure video entertainment, specifically in a peer-to-peer environment. / AFRIKAANSE OPSOMMING: Die wyse waarvolgens mense video verbruik is aan die verander met die ingebruikneming van nuwe tegnologie soos tabletrekenaars en slim televisiestelle. Hierdie nuwe tegnologie tesame met die Internet maak dat die verspreiding van video al hoe minder plaasvind deur middel van satellietuitsendings en al hoe meer versprei word deur die Internet. Aanlyn-Internetdienste bied deesdae dieselfde inhoud aan as wat voorheen slegs deur beeldsending versprei is. Hierdie dienste is egter slegs lewensvatbaar in lande met hoëspoed- en goedkoop Internetbandwydte. Daar is dus ’n behoefte aan alternatiewe tot hierdie dienste in lande waar bandwydte steeds duur en stadig is. Baie lande in Afrika kan in hierdie kategorie ingesluit word. In hierdie tesis word ’n videoverspreidingsplatform ontwerp en ontwikkel, wat van portuurnetwerke gebruik maak om hoëkwaliteit-beeldmateriaal te versprei. Die stelsel gebruik ’n bestaande portuurnetwerk-datavloeiprotokol as die premêre verspreidingsmeganisme, maar laat gebruikers ook toe om videoinhoud direk met ander gebruikers en dienste te deel. BitTorrent, DC++ en gebruikers wat hardeskywe met mekaar deel word hierby ingesluit. Ten einde die videoinhoud te beskerm ontwerp en implimenteer ons ’n sekuriteitstelsel wat verhoed dat gebruikers die videoinhoud onregmatig kan toe-eien, maar wat terselfdertyd die verspreiding van die data vergemaklik. Hierdie sluit die ontwikkeling van ’n pasgemaakte videospeler in. Die kern van die sekuriteitstelsel benodig ’n lae-bandwydte-Internetverbinding na ’n bediener wat sleutels uitsaai om die videoinhoud te ontsluit. Alhoewel nie daartoe beperk nie, is die platform gemik op hoëspoed-plaaslikegebiedsnetwerke met gratis bandwydte. Om die platvorm aan ’n haalbare sakemodel te laat voldoen het ons vir addisionele dienste soos videokatalogisering met soekfunksies, videoverbruikersmonitering en platvormadministrasie voorsiening gemaak. Die tesis sluit ’n literatuurstudie oor tegnieke en oplossings vir die beskerming van video data, spesifiek in die portuurnetwerke omgeving, in.

Data driven organizations gather information on various aspects of their endeavours and analyze that information to gain valuable insights or to increase automatization. Today, these organizations can choose from a wealth of specialized analytical libraries and platforms to meet their functional and non-functional requirements. Indeed, many common application scenarios involve the combination of multiple such libraries and platforms in order to provide a holistic perspective. Due to the scattered landscape of specialized analytical tools, this integration can result in complex and hard to evolve applications. In addition, the necessary movement of data between tools and formats can introduce a serious performance penalty. In this article we present a unified programming environment for analytical applications. The environment includes AL, a programming language that combines concepts of various common analytical domains. Further, the environment also includes a flexible compilation system that uses a language-, domain-, and platform independent program intermediate representation to separate high level application logic and physical organisation. We provide a detailed introduction of AL, establish our program intermediate representation as a generally useful abstraction, and give a detailed explanation of the translation of AL programs into workloads for our experimental shared-memory processing engine.