Hi Ruoshi,
There are several approaches to real-time simulation, including Simulink Real-Time and Simulink Desktop Real-Time. Which one is best for your project will depend on your requirements and hardware available.
As to the sequencing of your CAN messages, maybe a modelling approach based on enabled subsystems containing your "CAN Transmit" blocks would be the easiest to implement. For example, you could build a state machine running at a higher frequency (e.g., every 0.5 second) that generates 3 pulses based on the current simulation time value, and use each of those pulses to enable the desired subsystem at the right time. Does this make sense?
Best,
Dimitri