Dissertationen zum Thema „Efficient software implementation“

Software Defined Radio (SDR) is a relatively new approach to develop wireless communication systems. SDR presents a framework for developing flexible, reconfigurable devices intended to alleviate some of the issues arising from the evolution of wireless technology. The Software Communications Architecture (SCA), developed by the Joint Tactical Radio System program of the US Department of Defense, is an open architecture for implementing SDR, relying on commercial technology, standard interfaces, and well-known design patterns. Although the SCA is intended to provide easier, faster development of flexible applications that are upgradeable and maintainable, the acceptance of the architecture has been limited in part by traditional radio engineers' lack of understanding modern software engineering techniques. Because of the steep learning curve, some developers face frustration and serious delays when first introduced to the SCA. This work presents a comprehensive tutorial which introduces radio engineers to the SCA and the techniques used in it. Another concern for accepting the SCA are the performance, size, cost, and power consuption difficulties faced in early implementations of the architecture. Traditionally, SCA implementations have been developed for platforms based on General Purpose Processors. This approach, while believed to be the easiest to implement, does not make the best out of available processor technology. In order to provide a more efficient implementation of radios based on the SCA, we present the design and development of an SCA Core Framework version 2.2 for a homogeneous TI C64 DSP platform. This framework is implemented by leveraging the existing implementation of the Open-Source SCA Implementation::Embedded (OSSIE) by porting it to the C64 platform. Two sample waveforms are developed and deployed to demonstrate the functionality of the framework. Preliminary performance and memory footprint profiling results are provided.
Master of Science

Current cryptographic algorithms pose high standards of security yet they are susceptible to side-channel analysis (SCA). When it comes to implementation, the hardness of cryptography dangles on the weak link of side-channel information leakage. The widely adopted AES encryption algorithm, and others, can be easily broken when they are implemented without any resistance to SCA. This work applies state of the art techniques, namely Secret Sharing and Secure Multiparty Computation (SMC), on AES-128 encryption as a countermeasure to those attacks. This embedded C implementation explores multiple time-memory trade-offs for the design of its fundamental components, SMC and field arithmetic, to meet a variety of execution and storage demands. The performance and leakage assessment of this implementation for an ARM based micro-controller demonstrate the capabilities of masking schemes and prove their feasibility on embedded software.

Thesis (Ph.D.)--Case Western Reserve University, 2009
Department of Electrical Engineering Abstract Title from OhioLINK abstract screen (viewed on 10 April 2009) Available online via the OhioLINK ETD Center

Thesis (M.S. in Electrical Engineering and M.S. in Applied Mathematics)--Naval Postgraduate School, December 2008.
Thesis Advisor(s): Cristi, Roberto ; Rasmussen, Craig W. "December 2008." Description based on title screen as viewed on January 29, 2009. Includes bibliographical references (p. 61). Also available in print.

Real time convolution has many applications among others simulating room reverberation in audio processing. Non-uniformly partitioning filters could satisfy the both desired features of having a low latency and less computational complexity for an efficient convolution. However, distributing the computation to have an uniform demand on Central Processing Unit (CPU) is still challenging. Moreover, computational cost for very long filters is still not acceptable. In this thesis, a new algorithm is presented by taking advantage of the broad memory on Graphics Processing Units (GPU). Performing the computations of a non-uniformly partitioned block convolution on GPU could solve the problem of work load on CPU. It is shown that the computational time in this algorithm reduces for the filters with long length.

Dans cette thèse, nous étudions la conception de cryptosystèmes multivariés qui sont résistants contre les ordinateurs classiques et quantiques. En particulier, nous proposons deux schémas de signature digitale que j'ai soumis au processus de standardisation de cryptographie post-quantique du NIST : GeMSS et DualModeMS. Ces schémas sont basés sur la famille HFE. Nous proposons des paramètres de sécurité basés sur un état de l'art de vingt ans de cryptanalyse intensive. Puis, nous sélectionnons des paramètres qui maximisent l'efficacité. Nous la mesurons avec une nouvelle bibliothèque : MQsoft. MQsoft est une bibliothèque efficace en C qui supporte un large ensemble de paramètres pour les schémas basés sur HFE. Sa performance surpasse toutes les bibliothèques existantes. Nous expliquons dans cette thèse comment nous obtenons une telle performance. Tandis que GeMSS est un schéma qui a une grande clé publique, mais une signature très courte, DualModeMS est basé sur une transformation qui inverse ce comportement. Cependant, sa sécurité est basée sur l'hypothèse que le problème AMQ est difficile. Nous démontrons que cette hypothèse est vérifiée, et nous confirmons nos résultats avec des expériences utilisant les bases de Gröbner. Finalement, nous proposons la première implémentation de DualModeMS. Nous étudions comment l'implémenter efficacement, et comment optimiser le choix des paramètres de sécurité. Nous étendons aussi DualModeMS à l'utilisation du cryptosystème Rainbow à la place de HFE. Ceci permet d'obtenir des tailles de clés et signature intéressantes
In this thesis, we study the design of multivariate cryptosystems, which are resistant against classical and quantum computers. In particular, we study two digital signature schemes that I submitted to the NIST Post-Quantum Cryptography standardization process: GeMSS and DualModeMS. These schemes are based on the HFE family. We propose security parameters based on a state-of-the-art of twenty years of intensive cryptanalysis. Then, we select secure parameters which maximize the practical efficiency. We measure this with a new library: MQsoft. MQsoft is a fast library in C which supports a large set of parameters for HFE-based schemes. Its performance outperforms all existing libraries. We explain in this thesis how we obtain this result. Whereas GeMSS is a scheme which has a large public-key but a very short signature, DualModeMS is based on a transformation inverting this trade-off. However, its security is based on the assumption that the AMQ problem is hard. We demonstrate that this assumption is verified, and we confirm our results with experiences using Gröbner basis. Finally, we propose the first implementation of DualModeMS. We study how to implement it efficiently, as well as how to optimize the choice of security parameters. We also extend DualModeMS to the Rainbow cryptosystem instead of HFE. This allows to have interesting key sizes and signature sizes

The purpose of this master thesis was to investigate which of the two methods, resolution and tableaux, that is the most appropriate for automated theorem proving. This was done by implementing an automated theorem prover, comparing and documenting implementation problems, and measuring proving efficiency. In this thesis, I conclude that the resolution method might be more suitable for an automated theorem prover than tableaux, in the aspect of ease of implementation. Regarding the efficiency, the test results indicate that resolution is the better choice.
Syftet med detta magisterarbete var att undersöka vilken av de två metoderna, resolution och tablå, som är mest lämpad för automatisk teorembevisning. Detta gjordes genom att implementera en automatisk teorembevisare, jämföra och dokumentera problem, samt att mäta prestanda för bevisning. I detta arbete drar jag slutsatsen att resolutionsmetoden förmodligen är mer lämpad än tablåmetoden för en automatisk teorembevisare, med avseende på hur svår den är att implementera. När det gäller prestanda indikerar utförda tester att resolutionsmetoden är det bästa valet.

Sandvik Rock Tools is operating in a highly competitive market. This requires them to constantly improve the efficiency of their organization, in order for them to maintain their position as market leaders, both in a short-term and a long-term perspective. Sandvik Rock Tools is currently working on developing and further improving their supply chain. The supply chain is a key organizational function and improvements of the supply chain aim to yield benefits like lowered total costs and better utilized inventory levels. The purpose of our work has been to develop the supply chain between Sandvik’s in-market warehouse and their customer site at Dannemora. The main focus of this thesis has been to enable automatic data collection and to lower the stock levels at the customer site in Dannemora. This would be accomplished by improving the information flow between the two entities in the observed supply chain and by enabling real-time inventory information, easily accessible to the warehouse manager. A barcode system was implemented in the warehouse at the Dannemora customer site and was integrated with warehouse management software that was developed by the authors. The implemented barcode system and the warehouse management software were used at the Dannemora customer site for a period of two months. The effects of the implemented barcode system working in combination with the warehouse management software were observed in several operational areas after implementation, and the results were positive and beneficial. Firstly, the barcode system and warehouse management software simplified the warehouse manager’s decision process, by providing easy access to necessary warehouse management information, like realtime inventory information. Secondly, it made the day-to-day work routine of the warehouse manager easier, this being the case because the barcode system that had been implemented had digitalized and automated many tasks that had previously been performed manually by warehouse manager. An excellent example would be the simplified process for order placement and stocktaking. Thirdly, it improved the information flow between the Dannemora customer site and Sandvik’s in-market warehouse in Sandviken, thereby improving the overall effectiveness of the entire supply chain, through shared information. The barcode implementation successfully lowered inventory levels by six percent on average.

The master thesis deals with analysis and assessment of information system for high school using selected methods. Introduces proposal of solution with regard to security, reliability, functionality and financial aspects of the system. Final proposal should lead to improvement of ICT state in the analyzed institution.

碩士
國立中興大學
資訊科學系所
95
The Twenty First Century is an era comprised by the technological information societies. It is also the beginning of the Internet Era, because the Internet population is growing rapidly. Information technology evokes a revolution of learning along with the changes in learning styles. The coming of the Internet Era diversifies how humans acquire knowledge and the ways humans interact with one another. Web 2.0 creates another period of time when everyone participates, shares among with other, breaks set rules and then gains mutual benefits. The users from current society have already crossed the boundary from Web1.0 and are making their way into Web2.0 And so is the Education system. Nevertheless, the biggest difference between the concepts of Web2.0 and Web1.0 is that Web2.0 includes the ideas of teamwork and mutual sharing. As to the changes in the aspects of education and learning styles, students convert their passive learning habits to a more proactive approach towards knowledge acquisition and learning. The relationship between teachers and students has also become transformed from a top-down one-way direction to a more equally beneficial horizontal platform. F2blog is an open source software. It was developed by using the PHP5 language MYSQL5 database. It utilizes a BLOG which can be accessed by many users. F2blog adopts the Ajax technology. It reduces the waiting time for reloading the page. It provides perfect content management system. The plug-in function even allows the administrator to develop his own program which is very cost-effective. This study focuses on using the Web2.0 concept and F2blog open source software to construct a more effective platform for classroom network learning community. This platform helps teachers to manage their class, to provide on-line learning assistance, and to extend learning beyond the classroom. The purpose is to use the existing open source F2blog to construct a platform for classroom network learning community. At the same time, a service satisfaction analysis is conducted through PZB. To enhance the service satisfaction rate, modifications on system functionality are acted upon through the analysis. The study targets on IVY junior high school students who have participated in the practical community. The research results illustrate the increased willingness of participating in the network learning community. Teachers, parents and students are more motivated to participate in the communities. Therefore, the classroom network learning community created in this study is indeed practical and should be recommended and utilized by educational institutes.

The growing interest in video coding applications led to the development of video coding standards. Several successful standards have emerged, such as the ITU-T H.261, H.263, ISO/EEC MPEG-1 and MPEG-2. ITU-T recently announced its latest low bit rate video coding standard, the H.263 Version 2, also known as the H.263+ standard [25]. While this new low bit rate video coding standard provides better compression performance levels than those of the former ones, it is more complex and is more computationally, demanding. In this thesis, we develop coding and mapping algorithms that improve the video coding speed performance, making a reality the software implementation of interactive real-time low bit rate video coding on the Pentium MMX general purpose processor. First, using statistical properties of low resolution and slowly varying video sequences, we manage to significantly reduce the computation times of the most computationally intensive components of video coding, particularly the DCT, the IDCT, quantization and half pixel motion search. We also map some of the SIMD (Single Instruction Multiple Data) oriented functions onto Intel's M M X architecture. The developed algorithms are implemented using our public-domain H.263+ encoder/decoder software [45]. After the algorithmic optimization and the M M X mapping, the resulting H.263+ video encoder implementation runs approximately 3 times faster than the original unoptimized public-domain encoder implementation. Moreover, our optimized H.263+ compliant video coder implementation can encode/decode more than 15 frames of a QCIF (176x144) video sequence per second without sacrificing video reproduction quality.