Why is Model-Based Design Important in Embedded Systems?

Why and how to adopt Model-based Design?

The model-based design framework typically includes the following steps: (see figure 1)

Modeling: System modeling activities involve creating a mathematical and behavioral representation of the embedded system under consideration. It refers to a visual method for designing complex control systems, communication systems, and signal processing systems within an MBD framework.

Simulation: With MBD continuous testing as algorithms and real-time computational models are created and refined. Simulation is an alternative to building hardware prototype for testing purpose. During simulation, continuous-time systems are solved using numerical integration. Ideally, two types of solvers are used within MBD environments. These are:Fixed-step solvers: Fixed-step solvers use explicit methods to compute the next continuous state at fixed periodic intervals of time without any risk to infrastructure.Variable-step solvers: Variable-step solvers and continuous-time systems do not lend themselves well to deterministic real-time executables. So, this combination should be used carefully on those portions of the model that are targeted while the designing of embedded system software and subsequent code generation.

Rapid prototyping: MBD provides a rapid prototyping method of a product. It is a fast and cost-effective way for engineers to control signal processing, verify design at the early stage, and evaluate design trade-offs.

Embedded deployment: After rapid prototyping, a detailed software design activity is performed to convert the controller model to a detailed, executable software specification. Embedded code (often highly optimized) is then generated from the model for the detailed controller model and downloaded to the actual embedded microprocessor or ECU as part of the production software build.

In-the-loop testing: To combine hardware and production code into model-based testing, one can compare dynamic outputs of models with data collected through software-in-the-loop and processor-in-the-loop test or with data measured in the test lab, using the data inspector or logging tools.

Development flow of the Model-based Design:

Functional Modeling: In this, a model describes various functionalities of subsystems of embedded devices. It describes functions with few iterations and less effort in product development and ensures compatibility with other tools and processes. It follows an algorithm that defines electronic design rules of various process flows and components proposed for the embedded software.

Implementation Modeling: This process describes the control algorithm, components, sub-systems, and the environment in which the embedded device will be operated, using analysis techniques like simulation and it helps in the reduction of implementation complexity with better quality at the concept level and optimized code. It maintains the device component interactions about timing delays, continuous parameters and checks their compatibility across the entire embedded systems.

In brief, MBD provides graphical modeling environments consisting of block diagrams and state machines and is used to analyze, simulate, prototype, specify, and deploy software algorithms within a variety of embedded systems and applications, which is closer to real-world implementation.

How does a Model-based Design work in 1D and 3D simulations?

In the early stage of analysis of MBD simulations, design teams have little more than basic equations, diagrams, and constants for embedded software design lifecycle, which is known as 1D analysis. In the 1D analysis, design teams get a brief of direction, flexibility and rapid prototyping in order to access the design space. "1D simulation models usually possess a higher degree of abstraction and require a small amount of input data (compared to 3d models)," said , senior vice president for math and systems at Altair.

The above image showcases the upgrading of 1D simulation over 3D simulation. It shows that all the sub-systems combined with Design-of-Experiment techniques can be used upfront to generate parametric 1D-subsystem models. They can be even employed in the detailed design phase for subsystem optimization. If you are interested in combining both the worlds, read this , to know in detail.

In 1D, the design team learns more about its product, which becomes easier for design engineers to start the 3D simulations to . 3D is considered as a detailed simulation, which looks into the subsystem optimization of product design. Due to the high degree of abstraction, design team lacks accuracy in 1D. To overcome this, a 3D simulation model is being used for quantitative and detailed design phase in an embedded application.

Model-Based Design Tools: For Modeling & Simulation

Nowadays, to improve product quality and time to market or even greater design flexibility, engineers follow standard model-base1d designs like with MATLAB and .

These tools optimize and improve embedded design and development efficiency, system functionality and minimize overall design time in a simulation environment.

With the help of these MBD tools, engineers can: Streamline testing and verification workflows; Accelerate product development and Reduce development time

Tool for modeling

Simulink® developed by MathWorks is a graphical programming environment for modeling, simulation, and analysis. Its primary interface is to provide graphical block diagrams that allow engineers to draw models and customizable set of libraries that include a sink, source, linear and nonlinear component, and connector blocks.

Tool for simulation

Commands are necessary for running a group of simulation. Matlab® is a mathematical integration method for efficient technical computing, where problems and solutions are expressed in mathematical notations. It integrates computation, visualization, and programming in an easy-to-use environment.

Model-Based Design Benefits

The main benefit of using MBD is the auto-generation of code, which can eliminate human errors and allow reusability of code. In addition, by adopting model-based design incrementally, companies have consistently achieved immediate and tangible results like:

1. Faster time to the first demonstration

2. Faster time to market with a qualitative product

3. Quick turnaround of iterations without the need for hardware

4. Expanded capacity for handling complex embedded systems

5. Disciplined analysis, design and continuous testing in order to improve development effectiveness

6. Reusable models can improve development time and cost.

With an incremental approach, engineers can smoothly adopt model-based design and perform at even higher levels of speed, competence, and design quality in embedded systems.

eInfochips can help you optimize your model-based design process. We have accomplished a complete model-based application development for onboard fault diagnostics, electronic flight instrument system and display systems with deep expertise in a model-based design framework that includes MathWorks components like Stateflow, Simulink verification and validation tools, Polyspace Static Analysis, and Model advisor.

This blog is originally published at eInfochips.com - www.einfochips.com/blog/why-is-model-based-design-important-in-embedded-systems/