Rapid device-miniaturization keeps on inducing challenges in building energy efficient microprocessors. As the size of the transistors continuously decreasing, more uncertainties emerge in their operations. On the other hand, integrating more and more transistors on a single chip accentuates the need to lower its supply-voltage. This dissertation investigates one of the primary device uncertainties - timing error, in microprocessor performance bottleneck in NTC era. Then it proposes various innovative techniques to exploit these opportunities to maintain processor energy efficiency, in the context of emerging challenges. Evaluated with the cross-layer methodology, the proposed approaches achieve substantial improvements in processor energy efficiency, compared to other start-of-art techniques.

In this thesis, the analysis of the method of fundamental solution(MFS) is expanded for biharmonic equations. The bounds of errors are derived for the traditional and the Almansi's approaches in bounded simply-connected domains. The exponential and the polynomial convergence rates are obtained from highly and finite smooth solutions, respectively. Also the bounds of condition number are derived for the disk domains, to show the exponential growth rates. The analysis in this thesis is the first time to provide the rigor analysis of the CTM for biharmonic equations, and the intrinsic nature of accuracy and stability is similar to that of Laplace's equation. Numerical experiment are carried out for both smooth and singularity problems. The numerical results coincide with the theoretical analysis made. When the particular solutions satisfying the biharmonic equation can be found, the method of particular solutions(MPS) is always superior to MFS, supported by numerical examples. However, if such singular particular solutions near the singular points can not be found, the local refinement of collocation nodes and the greedy adaptive techniques can be used. It seems that the greedy adaptive techniques may provide a better solution for singularity problems. Beside, the numerical solutions by Almansi's approaches are slightly better in accuracy and stability than those by the traditional FS. Hence, the MFS with Almansi's approaches is recommended, due to the simple analysis, which can be obtained directly from the analysis of MFS for Laplace's equation.

In this dissertation we investigate the applicability of different adaptive techniques to improve the effectiveness and efficiency of the regression testing. Initially, we introduce the concept of regression testing. We then perform a literature review of current practices and state-of-the-art regression testing techniques. Finally, we advance the regression testing techniques by performing four empirical studies in which we use different types of information (e.g. user session, source code, code commit, etc.) to investigate the effectiveness of each software metric on fault detection capability for different software environments. In our first empirical study, we show the effectiveness of applying user session information for test case prioritization. In our next study, we apply learning from the previous study, and implement a collaborative filtering recommender system for test case prioritization, which uses user sessions and change history information as input parameter, and return the risk score associated with each component. Results of this study show that our recommender system improves the effectiveness of test prioritization; the performance of our approach was particularly noteworthy when we were under time constraints. We then investigate the merits of multi-objective testing over single objective techniques with a graph-based testing framework. Results of this study indicate that the use of the graph-based technique reduces the algorithm execution time considerably, while being just as effective as the greedy algorithms in terms of fault detection rate. Finally, we apply the knowledge from the previous studies and implement a query answering framework for regression test selection. This framework is built based on a graph database and uses fault history information and test diversity in attempt to select the most effective set of test cases in term of fault detection capability. Our empirical evaluation of this study with four open source programs shows that our approach can be effective and efficient by selecting a far smaller subset of tests compared to the existing techniques.

<p>Free-space optical (FSO) communication has witnessed rapid development recently in response to ever-increasing demands for greater bandwidth. FSO links provide fiberspeed with the flexibility of wireless. Commercially available systems offer transmission speeds up to 2.5 Gbps, 5 Gbps and 10 Gbps, and demonstration systems report data rates as high as 160 Gbps. Its advantages also include license free operation, high immunity to interference, and ease of deployment. However, FSO systems are sensitive to adverse weather conditions such as fog, rain and snow.</p> <p>In order to improve the availability of FSO channels degraded by atmospheric turbulence and varying weather conditions, the effects of channel gain variations must be compensated. In this thesis, two rate-adaptive techniques, punctured low-density parity-check (LDPC) codes and Raptor codes, are studied using experimental data measured over a1.87 km terrestrial FSO link.</p> <p>Rate-adaptive performances with punctured LDPC codes and Raptor codes are evaluated in terms of outage probability and throughput. In comparison to uncoded system, rate-adaptive systems with both techniques demonstrate significant improvement of throughput and mitigation of outage probability especially in rainy weather. Due to its flexible rate-adaptation and simple hardware implementation, Raptor coded systems are judged more applicable to be implemented in field-programmable gate array (FPGA) based hardware. A dedicated decoding structure is proposed and tested, showing remarkable improvement in resource efficiency as compared to traditional Gauss-Jordan (GJ) decoding structures.</p> / Master of Applied Science (MASc)