DSP Blockset

What Is the DSP Blockset?

The DSP Blockset is a collection of block libraries for use with the Simulink dynamic system simulation environment.

The DSP Blockset libraries are designed specifically for digital signal processing (DSP) applications, and include key operations such as classical, multirate, and adaptive filtering, matrix manipulation and linear algebra, statistics, time-frequency transforms, and more.

Key Features

The DSP Blockset extends the Simulink environment by providing core components and algorithms for DSP systems. You can use blocks from the DSP Blockset in the same way that you would use any other Simulink blocks, combining them with blocks from other libraries to create sophisticated DSP systems.

A few of the important features are described in the following sections:

• Frame-Based Operations
• Matrix Support
• Adaptive and Multirate Filtering
• Statistical Operations
• Linear Algebra
• Parametric Estimation
• Real-Time Code Generation

Frame-Based Operations

Most real-time DSP systems optimize throughput rates by processing data in "batch" or "frame-based" mode, where each batch or frame is a collection of consecutive signal samples that have been buffered into a single unit. By propagating these multisample frames instead of the individual signal samples, the DSP system can best take advantage of the speed of DSP algorithm execution, while simultaneously reducing the demands placed on the data acquisition (DAQ) hardware.

The DSP Blockset delivers this same high level of performance for both simulation and code generation by incorporating frame-processing capability into all of its blocks. A completely frame-based model can run several times faster than the same model processing sample-by-sample; faster still if data sources are frame based.

See Sample Rates and Frame Rates for more information.

Matrix Support

The DSP Blockset takes full advantage of the matrix format of Simulink. Some typical uses of matrices in DSP simulations are

• General two-dimensional array

• A matrix can be used in its traditional mathematical capacity, as a simple structured array of numbers. Most blocks for general matrix operations are found in the Matrices and Linear Algebra library.

• Factored submatrices

• A number of the matrix factorization blocks in the Matrix Factorizations library store the submatrix factors (i.e., lower and upper submatrices) in a single compound matrix. See the LDL Factorization and LU Factorization blocks for examples.

• Multichannel frame-based signal

• The standard format for multichannel frame-based data is a matrix containing each channel's data in a separate column. A matrix with three columns, for example, contains three channels of data, one frame per channel. The number of rows in such a matrix is the number of samples in each frame.

See the following sections for more information about working with matrices:

• Multichannel Signals
• Creating Signals
• Constructing Signals
• Importing Signals

Adaptive and Multirate Filtering

The Adaptive Filters and Multirate Filters libraries provide key tools for the construction of advanced DSP systems. Adaptive filter blocks are parameterized to support the rapid tailoring of DSP algorithms to application-specific environments, and effortless "what if" experimentation. The multirate filtering algorithms employ polyphase implementations for efficient simulation and real-time code execution.

See Multirate Filters and Adaptive Filters for more information.

Statistical Operations

Use the blocks in the Statistics library for basic statistical analysis. These blocks calculate measures of central tendency and spread (e.g., mean, standard deviation, and so on), as well as the frequency distribution of input values (histograms).

See Statistics for more information.

Linear Algebra

The Matrices and Linear Algebra library provides a wide variety of matrix factorization methods, and equation solvers based on these methods. The popular Cholesky, LU, LDL, and QR factorizations are all available.

See Linear Algebra for more information.

Parametric Estimation

The Parametric Estimation library provides a number of methods for modeling a signal as the output of an AR system. The methods include the Burg AR Estimator, Covariance AR Estimator, Modified Covariance AR Estimator, and Yule-Walker AR Estimator, which allow you to compute the AR system parameters based on forward error minimization, backward error minimization, or both.

Real-Time Code Generation

You can also use the separate Real-Time Workshop product to generate optimized, compact, ANSI C code for models containing blocks from the DSP Blockset.

See Code Generation Support for more information.

Introduction

What Is in the DSP Blockset?