{"id":9190,"date":"2019-11-15T15:50:51","date_gmt":"2019-11-15T20:50:51","guid":{"rendered":"https:\/\/blogs.mathworks.com\/simulink\/?p=9190"},"modified":"2019-11-18T12:59:57","modified_gmt":"2019-11-18T17:59:57","slug":"new-in-r2019b-subsystem-reference","status":"publish","type":"post","link":"https:\/\/blogs.mathworks.com\/simulink\/2019\/11\/15\/new-in-r2019b-subsystem-reference\/","title":{"rendered":"New in R2019b: Subsystem Reference"},"content":{"rendered":"

Today I want to introduce my favorite new feature in MATLAB R2019b: Subsystem Reference<\/a><\/p>\n

Let's see what is it, how it works, and when you should be using it.<\/p>\n

Creating a Subsystem File<\/strong><\/p>\n

In R2019b, we are introducing a new type of block diagram. In addition to model files and library files, it is now possible to create subsystem files:<\/p>\n

\"New<\/p>\n

In the Current Folder window, you can differentiate models, subsystems and libraries by their icon and type:<\/p>\n

\"Compare<\/p>\n

From the command-line, if the subsystem is open you can determine what type of block diagram it is using get_param:<\/p>\n

\"File<\/p>\n

If you need to determine which type of block diagram an slx-file is without opening it or even loading it in memory, you can also use Simulink.MDLInfo:<\/p>\n

\"Block<\/p>\n

Referencing a Subsystem File<\/strong><\/p>\n

The first important thing to realize is that a subsystem file cannot be simulated by itself. If you want to simulate or generate code for the block diagram stored in a subsystem file, you need to reference it in a model.<\/p>\n

To reference a subsystem file, you start with a Subsystem block<\/a>. In R2019b, you will notice that the Subsystem block has a new tab named Subsystem Reference. For convenience, we also added to the Simulink Library a block named Subsystem Reference, which is simply a Subsystem block pre-configured to reference a subsystem file.<\/p>\n

Here is an animation where I configure and edit a referenced subsystem. As you can see, entering the subsystem and editing it is as simple and easy as editing a standard subsystem.<\/p>\n

\"Editing<\/p>\n

Converting an Existing Subsystem<\/strong><\/p>\n

If you have an existing model and would like to convert existing subsystems to referenced subsystems, you can simply right-click on the existing subsystem to do the conversion.<\/p>\n

\"Converting<\/p>\n

Instance Specific Context Awareness<\/strong><\/p>\n

In a way similar to library blocks, it is possible to use multiple instances of the same subsystem in a model, and each instance adapts to its context. The next image shows a model with 3 instances of the same subsystem, one receiving a complex scalar double, one receiving a matrix of 2x3 and the third instance receiving a vector of two integers.<\/p>\n

\"Converting<\/p>\n

Masking<\/strong><\/p>\n

If you want multiple instances of a referenced subsystem to take different parameter values, you can create a mask for the subsystem. This is done from the Modeling tab:<\/p>\n

\"Masking<\/p>\n

This will allow you to pass different parameter values to different instances of the subsystem. The next image shows 3 instances of the same subsystem. The two first take different values for a Gain block, and the last one takes a vector of 2 as input.<\/p>\n

\"Multiple<\/p>\n

When To Use Subsystem Reference?<\/strong><\/p>\n

Now the big question: When to use subsystem reference?<\/strong><\/p>\n

The official answer to this question is to read the documentation section Component-Based Modeling Guidelines<\/a>. In this section of the documentation, you will find this flow chart:<\/p>\n

\"Componentization<\/p>\n

Here is my condensed version of how I make this decision. To componentize a model, you need to choose between 3 technologies: libraries, model reference and subsystem reference:<\/p>\n