Graphical Modeling Fixed-Point Signal Visualization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In graphical modeling environments, optimizing the representation of signals and block parameters using fixed-point data types is crucial for reducing code size and increasing execution speed, but existing tools lack effective methods for determining the appropriate range and precision, leading to potential overflows and underflows.
Innovation Solution
A graphical modeling environment that allows users to represent signals and block parameters in fixed-point data types, providing visual cues such as coloring and shading to indicate maximum and minimum values, overflow/underflow occurrences, and bit allocation, enabling users to adjust scaling factors and refine the model design.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If fixed-point data types are used in graphical models, then code size is reduced and execution speed is increased, but determining appropriate range and precision becomes difficult leading to overflows and underflows
Solution Approach 1:
The system automatically analyzes signal statistics during simulation and provides feedback information about maximum/minimum values, overflow/underflow occurrences, and bit allocation adequacy. This feedback loop enables users to iteratively refine their fixed-point designs by adjusting scaling factors based on actual signal behavior observed during simulation, thereby preventing overflows and underflows while maintaining execution efficiency.
Solution Approach 2:
The system performs preliminary analysis of signal characteristics before final code generation. By simulating the model first and collecting statistical information about signal ranges and distributions, the system prepares the necessary data to determine optimal fixed-point parameters in advance, allowing users to set appropriate word lengths and fraction lengths before deploying to hardware, thus preventing runtime overflows and underflows.
2Ease of manufacture
If fixed-point data types are used in graphical models, then hardware cost is reduced, but determining appropriate bit allocation becomes complex
Solution Approach 1:
The system automatically performs the complex task of analyzing signal statistics and determining optimal bit allocation without requiring manual intervention. It self-services by collecting data during simulation, computing maximum/minimum values and overflow/underflow counts, and presenting analyzed results that guide users in setting appropriate fixed-point parameters, thereby simplifying what would otherwise be a complex manual process.
Solution Approach 2:
The system acts as an intermediary between the graphical model and the fixed-point code generation process. It introduces an intermediate analysis step that bridges the gap between model simulation and hardware deployment, providing users with processed information about signal characteristics and bit allocation recommendations, thus mediating the complexity of determining optimal fixed-point parameters for cost-effective hardware implementation.
3Measurement precision
If more bits are allocated to fixed-point signals, then precision is improved, but code size increases and execution speed decreases
Solution Approach 1:
The system enables dynamic adjustment of fixed-point parameters (word length, fraction length, scaling factors) based on actual signal statistics observed during simulation. By analyzing the true range and distribution of signal values, users can optimize these parameters to achieve the minimum necessary precision for each signal, avoiding excessive bit allocation that would slow down execution while maintaining adequate precision through data-driven parameter selection.
Data Source
AI summary
A graphical modeling environment is provided for a user to build a model in which signals and/or block parameters can be represented using a fixed-point data type. When the graphical model is executed, the graphical model may generate fixed-point data type signals. An exemplary embodiment may information about fixed-point data type signals on the graphical model. Based on the information provided on the graphical model, the user may be able to determine whether an appropriate number of bits are allocated to the integer part of the fixed-point data type signals and/or whether the total number of bits, i.e., the word length is adequate or over-engineered.