- ホーム
- > 洋書
- > 英文書
- > Science / Mathematics
Full Description
The continuously increasing computing power of Digital Signal Processing makes it now possible to efficiently implement Non-linear Algorithms for Signal Processing (NLSP). This book proposes a comprehensive review of Non-Linear Signal Processing Methods and the associated Parameter Estimation principles. The various existing approaches are considered: Classical descriptions (Hammerstein models, Volterra Equations ?), and more modern ones like Neural Network based ones, Wavelet Transform based decompositions, etc. The estimation of parameters is also considered: Classical Kalman Filter, Particle Filtering, and Self Learning Networks.
Contents
1. Basic classification of Non Linear (NL) representations ofsignals: with or without memory 1.1 Memoryless systems effects on signals: Probability Densitytransformations. Random Processes Moment transformations: Pricetheorem and its generalizations. 1.2 Time Dependent NL signal models: integral and differentialequations (Fredholm, Volterra, etc.). 2. Modeling Non-Linear systems 2.1 Hammerstein separable Models 2.2 Cellular networks: Neural Networks, Support VectorMachines 2.3 State Space Equation based: Extended Kalman Filter 3. Parameter estimation in NL systems 3.1 Known Input Methods: Kalman, Least Squares and RecursiveLeast Squares, Supervised (i.e. 'with learning phase'): NeuralNetworks 3.2 Self-learning mode: Kohonen-like algorithms 4. Selected application examples derived from: 4.1 Basic Signal Processing: Polynomial NL systems,hard-limiters, clippers, etc. 4.2 Space Telecommunications: Satellite On-board Solid StatePower Amplifier, Non-Linear Channel Equalizers.
