Signal Processing Blockset™ 6.7

Product Description

• Introduction and Key Features
• Working with the Signal Processing Blockset
• Modeling and Simulation
• Using Floating-Point and Fixed-Point Data Types
• Generating and Optimizing C Code
• Validating and Verifying Models

Using Floating-Point and Fixed-Point Data Types

All blocks in the Signal Processing Blockset support double-precision and single-precision floating-point data types. Many also support integer and fixed-data point data types. (Fixed-point support requires Simulink Fixed Point, available separately.) The Signal Processing Blockset supports Boolean data types for appropriate (logical) signals and parameters.

Blocks in the Signal Processing Blockset operate in a homogeneous manner with respect to data types: floating-point inputs cause the block to operate in a purely floating-point mode and fixed-point inputs cause the block to operate in a purely fixed-point mode. In fixed-point mode, you can use the Fixed-Point Settings interface to track and report flows as well as minima and maxima of internal variables.



The fixed-point Acoustic Noise Canceler model (above) uses the LMS adaptive filter to automatically remove noise from the input signal. The waterfall plot shows amplitude over time. Click on image to see enlarged view.

Applying Extended Fixed-Point Capabilities

The Signal Processing Blockset provides dialogs that help you configure fixed-point attributes. The blockset includes automatic settings for most quantities. The Fixed-Point Settings GUI facilitates the conversion of floating-point models to fixed point. The GUI tracks overflows and maxima and minima, helping you to configure fixed-point models.

The Signal Processing Blockset provides extended fixed-point support, including:

• Word sizes for signals and parameters (simulation from 1 to 128 bits and code generation from 1 to 32 bits)
• Arbitrary binary-point placement
• Real or complex numbers for fixed-point quantities
• Signed and unsigned fixed-point quantities
• Wrap or saturation arithmetic for overflows
• Rounding or truncation

The Fixed-Point Settings GUI helps you migrate models from floating point to fixed point. The middle wave shows the error between double precision and fixed point. Click on image to see enlarged view.

The Signal Processing Blockset includes automatic configuration modes and support for configuring blocks for fixed-point operations. For example, you can specify the accumulator and multiplier size to ensure compatibility for a particular target chip, such as a digital signal processor. You can manually configure quantities to any format using the binary-point-scaling mode.

Examples of automatic configuration modes and fixed-point operations include the following:

• Binary point of a filter’s coefficient is automatically located based on user-defined word length, precision, and actual values.
• Product output retains all bits in the products between filter coefficients and input values, i.e., two 16-bit multiples equal a 32-bit product.
• Accumulator is configured to accumulate individual products and avoid overflows.

Using the binary-point-scaling mode, you can also manually configure quantities to any format.



A fixed-point model of the physical layer of ultra wideband (UWB) built with the Signal Processing Blockset. The model enables engineers to assess how fixed-point data types affect end-to-end system performance. Click on image to see enlarged view.

Designing and Modifying Filters

You can design digital filters using the Signal Processing Blockset with the Signal Processing Toolbox. The Filter Design and Analysis Tool in the Signal Processing Toolbox is integrated with Signal Processing Blockset blocks to let you design, analyze, and modify digital filter coefficients.

You can design fixed-point filters and advanced filter architectures for your simulations using the Filter Design Toolbox (available separately). The Signal Processing Blockset accepts filter parameters created with the Signal Processing Toolbox and the Filter Design Toolbox. You can use these parameters and the Filter Realization Wizard to generate custom filters in block-diagram format for use with the Signal Processing Blockset.