本文旨在将新兴的混沌、分形理论引入语音信号处理。本文提出了一种新的语音信号相空间重构方法，分析、统计了语音信号最大Lyapunov指数及分维度的分布，并提出了基于分形码本的语音信号码激励线性预测编码新算法。本文的研究表明，混沌、分形理论在语音信号处理中有良好的应用前景。

The propagation characteristics of magnetoplasma waveguide coated with a finite thickness of dielectric material are presented. Because the boundary conditions at all interfaces can be represented by the matrix equations, the characteristic equation can be obtained by successively using the matrix manipulations. The dependence relations of attenuation and dispersion of the fundamental mode with plasma frequency, collision frequency, and electron cyclotron frequency are shown.

A new noise-robust speecb tecognition method is presented based on the difference in the power spectrum. The idea is to remove tile additive noise by filtering in the power spectrum domain. Feature extraction is carried out in two steps: (i) tbe sbort-tilne power spectrum of the speech signal is allowed to pass through a filter bank; and (ii) the differences in the filter outputs are calculated. Theoretical analysis and experimental results show tbat using the proposed featurcs can significantly improve the recogniser's performance in a noisy environment.

A scheduling algorithm denoted frame-based approximate sorting fair queueing (FASFQ) is proposed. FASFQ assembles the packets of G-sessions (sessions with rate reservation) as frames which are transmitted one by one. The packets of E-sessions ('best-effort' sessions) are sent during the intervals between frames. FASFQ reduces the sorting complexity by replacing packet-sorting with frame-sorting, and can satisfy the different QoS requirements of different sessions. Analyses and simulations show that FASFQ can guarantee reserved rates for G-sessions and provide good isolation among sessions.

Based on the homotopy theory, this paper presents a robust nonlinear system identification algorithm. Theoretical analysis and experimental results show that, unlike the ordinary nonlinear system identification methods, the new method is more robust and easier to compute, has very good convergence property and compatibility with the traditional linear system identification.