logoCRL.png
 
Informationen zur Lehrveranstaltung:
 
Adaptive & Array Signal Processing
Wahlobligatorische Lehrveranstaltung für die Studiengänge

Weitere Interessenten sind herzlich willkommen.

Periodizität:

Vorlesender:

Ort und Zeit:

Vorlesungssprache:

Motivation:

This lecture provides an introduction to adaptive filters and array signal processing techniques. Most of the practical examples are taken from mobile communications. The lecture is given in English to familiarize the students with the linguistic requirements of a global economy, where technical discussions between international partners are usually conducted in English.

Homeworks:

Homework assignments will be given to you as the semester progresses. Most homework problems will be to show some results discussed in class. This will involve theoretical proofs as well as simulation studies using Matlab. In many cases, assignments can involve examining the performance of a system or extracting key parameters from measurements. Most of the topics covered by the course is best learned by working with the problems through a combination of computer simulations and theoretical analysis. The homework is therefore one of the most important parts of the course.

Cooperative group study on the homework is encouraged, but simply copying someone else's work is unethical and will leave you unprepared for exams. Significant insight can be gained by studying with another student or in a group, provided you discipline yourself to find your own solutions first before comparing results. Rely on other's help only when you have exhausted all of your own ideas or have made no progress for an extended period of time.

24.11.2017: new homework available.

Slides and files:

- complete: slides with the numbers used in the lectures
- compact: file without additional slides, therefore different numbers

Written Exam:

Sample exams:

Overview:

1 Introduction
    - Adaptive Filters
    - Single channel adaptive equalization (temporal filter)
    - Multi channel adaptive beamforming (spatial filter)
2 Mathematical Background
  2.1 Calculus
    - Gradients
    - Differentiation with respect to a complex vector
    - Quadratic optimization with linear constraints (method of Lagrangian multipliers)
  2.2 Stochastic processes
    - Stationary processes
    - Time averages
    - Ergodic processes
    - Correlation matrices
  2.3 Linear algebra
    - Eigenvalue decomposition
    - Eigenfilter
    - Linear system of equations
    - Four fundamental subspaces
    - Singular value decomposition
    - Generalized inverse of a matrix
    - Projections
    - Low rank modeling
3 Adaptive Filters
  3.1 Linear Optimum Filtering (Wiener Filters)
    - Principle of Orthogonality
    - Wiener-Hopf equations
    - Error-performance surface
    - MMSE (minimum mean-squared error)
    - Canonical form of the error-performance surface
    - MMSE filtering in case of linear Models
  3.2 Linearly Constrained Minimum Variance Filter
    - LCMV beamformer
    - Minimum Variance Distortionless Response (MVDR) spectrum: Capon's method
    - LCMV beamforming with multiple linear constraints
  3.3 Generalized Sidelobe Canceler
  3.4 Iterative Solution of the Normal Equations
    - Steepest descent algorithm
    - Stability of the algorithm
    - Optimization of the step-size
  3.5 Least Mean Square (LMS) Algorithm
  3.6 Recursive Least Squares (RLS) Algorithm
4 High-Resolution Parameter Estimation
    - Data model (DOA estimation)
    - Eigendecomposition of the spatial correlation matrix at the receive array
    - Subspace estimates
    - Estimation of the model order
  4.1 Spectral MUSIC
    - DOA estimation
    - Example: uniform linear array (ULA)
    - Root-MUSIC for ULAs
    - Periodogram
    - MVDR spatial spectrum estimation (review)
  4.2 Standard ESPRIT
    - Selection matrices
    - Shift invariance property
  4.3 Signal Reconstruction
    - LS solution
    - MVDR / BLUE solution
    - Wiener solution (MMSE solution)
    - Antenna patterns
  4.4 Spatial smoothing
  4.5 Forward-backward averaging
  4.6 Real-valued subspace estimation
  4.7 1-D Unitary ESPRIT
    - Reliability test
    - Applications in Audio Coding
  4.8 Multidimensional Extensions
    - 2-D MUSIC
    - 2-D Unitary ESPRIT
    - R-D Unitary ESPRIT
  4.9 Multidimensional Real-Time Channel Sounding
  4.10 Direction of Arrival Estimation with Hexagonal ESPAR Arrays
5 Tensor-Based Signal Processing
  5.1 Introduction and Motivation
  5.2 Fundamental Concepts of Tensor Algebra
  5.3 Elementary Tensor Decompositions
    - Higher Order SVD (HOSVD)
    - CANDECOMP / PARAFAC (CP) Decomposition
  5.4 Tensors in Selected Signal Processing Applications
6 Maximum Likelihood Estimators
  6.1 Maximum Likelihood Principle
  6.2 The Fisher Information Matrix and the Cramer Rao Lower Bound (CRLB)
    - Efficiency
    - CRLB for 1-D direction finding applications
    - Asymptotic CRLB

References:



last change 2017-11-30, Impressum

























Counter html Code