Lam Yuet Ming. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2003. / Includes bibliographical references (leaves 80-86). / Abstracts in English and Chinese. / Chapter 1 --- Introduction --- p.1 / Chapter 1.1 --- Motivation --- p.1 / Chapter 1.1.1 --- Difficulties of fixed-point design --- p.1 / Chapter 1.1.2 --- Why still fixed-point? --- p.2 / Chapter 1.1.3 --- Difficulties of converting floating-point to fixed-point --- p.2 / Chapter 1.1.4 --- Why wordlength optimization? --- p.3 / Chapter 1.2 --- Objectives --- p.3 / Chapter 1.3 --- Contributions --- p.3 / Chapter 1.4 --- Thesis Organization --- p.4 / Chapter 2 --- Review --- p.5 / Chapter 2.1 --- Introduction --- p.5 / Chapter 2.2 --- Simulation approach to address quantization issue --- p.6 / Chapter 2.3 --- Analytical approach to address quantization issue --- p.8 / Chapter 2.4 --- Implementation of speech systems --- p.9 / Chapter 2.5 --- Discussion --- p.10 / Chapter 2.6 --- Summary --- p.11 / Chapter 3 --- Fixed-point arithmetic background --- p.12 / Chapter 3.1 --- Introduction --- p.12 / Chapter 3.2 --- Fixed-point representation --- p.12 / Chapter 3.3 --- Fixed-point addition/subtraction --- p.14 / Chapter 3.4 --- Fixed-point multiplication --- p.16 / Chapter 3.5 --- Fixed-point division --- p.18 / Chapter 3.6 --- Summary --- p.20 / Chapter 4 --- Fixed-point class implementation --- p.21 / Chapter 4.1 --- Introduction --- p.21 / Chapter 4.2 --- Fixed-point simulation using overloading --- p.21 / Chapter 4.3 --- Fixed-point class implementation --- p.24 / Chapter 4.3.1 --- Fixed-point object declaration --- p.24 / Chapter 4.3.2 --- Overload the operators --- p.25 / Chapter 4.3.3 --- Arithmetic operations --- p.26 / Chapter 4.3.4 --- Automatic monitoring of dynamic range --- p.27 / Chapter 4.3.5 --- Automatic calculation of quantization error --- p.27 / Chapter 4.3.6 --- Array supporting --- p.28 / Chapter 4.3.7 --- Cosine calculation --- p.28 / Chapter 4.4 --- Summary --- p.29 / Chapter 5 --- Speech recognition background --- p.30 / Chapter 5.1 --- Introduction --- p.30 / Chapter 5.2 --- Isolated word recognition system overview --- p.30 / Chapter 5.3 --- Linear predictive coding processor --- p.32 / Chapter 5.3.1 --- The LPC model --- p.32 / Chapter 5.3.2 --- The LPC processor --- p.33 / Chapter 5.4 --- Vector quantization --- p.36 / Chapter 5.5 --- Hidden Markov model --- p.38 / Chapter 5.6 --- Summary --- p.40 / Chapter 6 --- Optimization --- p.41 / Chapter 6.1 --- Introduction --- p.41 / Chapter 6.2 --- Simplex Method --- p.41 / Chapter 6.2.1 --- Initialization --- p.42 / Chapter 6.2.2 --- Reflection --- p.42 / Chapter 6.2.3 --- Expansion --- p.44 / Chapter 6.2.4 --- Contraction --- p.44 / Chapter 6.2.5 --- Stop --- p.45 / Chapter 6.3 --- One-dimensional optimization approach --- p.45 / Chapter 6.3.1 --- One-dimensional optimization approach --- p.46 / Chapter 6.3.2 --- Search space reduction --- p.47 / Chapter 6.3.3 --- Speeding up convergence --- p.48 / Chapter 6.4 --- Summary --- p.50 / Chapter 7 --- Word Recognition System Design Methodology --- p.51 / Chapter 7.1 --- Introduction --- p.51 / Chapter 7.2 --- Framework design --- p.51 / Chapter 7.2.1 --- Fixed-point class --- p.52 / Chapter 7.2.2 --- Fixed-point application --- p.53 / Chapter 7.2.3 --- Optimizer --- p.53 / Chapter 7.3 --- Speech system implementation --- p.54 / Chapter 7.3.1 --- Model training --- p.54 / Chapter 7.3.2 --- Simulate the isolated word recognition system --- p.56 / Chapter 7.3.3 --- Hardware cost model --- p.57 / Chapter 7.3.4 --- Cost function --- p.58 / Chapter 7.3.5 --- Fraction size optimization --- p.59 / Chapter 7.3.6 --- One-dimensional optimization --- p.61 / Chapter 7.4 --- Summary --- p.63 / Chapter 8 --- Results --- p.64 / Chapter 8.1 --- Model training --- p.64 / Chapter 8.2 --- Simplex method optimization --- p.65 / Chapter 8.2.1 --- Simulation platform --- p.65 / Chapter 8.2.2 --- System level optimization --- p.66 / Chapter 8.2.3 --- LPC processor optimization --- p.67 / Chapter 8.2.4 --- One-dimensional optimization --- p.68 / Chapter 8.3 --- Speeding up the optimization convergence --- p.71 / Chapter 8.4 --- Optimization criteria --- p.73 / Chapter 8.5 --- Summary --- p.75 / Chapter 9 --- Conclusion --- p.76 / Chapter 9.1 --- Search space reduction --- p.76 / Chapter 9.2 --- Speeding up the searching --- p.77 / Chapter 9.3 --- Optimization criteria --- p.77 / Chapter 9.4 --- Flexibility of the framework design --- p.78 / Chapter 9.5 --- Further development --- p.78 / Bibliography --- p.80

published_or_final_version / Electrical Engineering / Master / Master of Philosophy

No description available.

No description available.

by Leung Pak Ming. / Thesis (M.Phil.)--Chinese University of Hong Kong, 1980. / Bibliography: leaves 199-206.

Blind Source Separation is a modern signal processing technique which recovers both the unknown sources and unknown mixing systems from only measured mixtures of signals. It has application in diverse fields such as communication, image processing, geological exploration and biomedical signal processing etc. This project studies the BSS problem, develop separation methods and reveal the potential for mechanical engineering applications. There are two models for blind source separation corresponding to the two ways that the sources are mixed, the instantaneous mixing model and the convolved mixing model. The author carried out a theoretical study of the first model by proposing an idea called Redundant Data Elimination which leads to geometric interpretation of the model, explains that circular distribution property is the reason why Gaussian signal mixtures can not be separated, and showed that this idea can improve separation accuracy for unsymmetrically distributed sources. This new idea enabled evaluation and comparison of two well-known algorithms and proposal of a simplified algorithm based on Joint Approximate Diagonalization of fourth order cumulant matrices, which is further developed by determining an optimized parameter value for separation convergence. Also based on the understanding from the RDE, an outlier spherical projection method is proposed to improve separation accuracy against outlier errors. Mechanical vibration or acoustic problems belong to the second model. After some theoretical study of the problem and the model, a novel application of the Blind Least Mean Square algorithm using Gray's variable norm as cost function is applied to engine vibration data to separate piston slap, fuel injection noise and cylinder pressure effects. Further, the algorithm is combined with a deflation algorithm for successive subtraction of recovered source responses from the measured mixture to enable the recovery of more sources. The algorithms are verified to be successful by simulation, and the separated engine sources are proved reasonable by analysing the engine operation and physical properties of the sources. The author also studied the relationship between these two models, the problems of different approaches for solving the model such as the frequency domain approach and the Bussgang approach, and sets out future research interests.

Not available / text

published_or_final_version / Computer Science / Doctoral / Doctor of Philosophy

No description available.

No description available.