Power and Thermal Modeling Approach for Embedded and Automotive using ESL Tools

Did you know that an S-class Mercedes Benz can use 100 microprocessor-based electronic control units (ECUs) networking throughout the vehicle that run 20-100 million lines of code (Source: IEEE)?

2014 Mercedes-Benz CLA

Here’s a quick list of all the places that you will find software controlling hardware in an automobile:

[TABLE] style=”width: 500px”
|-
| Air-bag system
| Antilock Brakes
| Automatic Transmission
|-
| Alarm system
| Climate control
| Collision-avoidance system
|-
| Cruise control
| Communication system
| Dashboard instrumentation
|-
| Electronic stability control
| Engine ignition
| Engine control
|-
| Electronic-set control
| Entertainment system
| Navigation system
|-
| Power steering
| Tire-pressure monitoring
| Windshield wiper control
|-

Automotive subsystems are highly modularized from a variety of qualified suppliers. The use of virtual platform (VP) and system level modeling has been embraced by the automotive industry to take a “systems engineering approach” both for prototyping and development, specifically in firmware and control systems.

Electronic components per auto will grow some 26 percent over the decade to 2022, from data gleaned by the Fuji Chimera Research Institute. For an electric vehicle like Model S from Tesla Motors, electronics tally about 50% of the car’s value compared to 20 percent on a gas-power car. Now that looks to me like a stable and lucrative market to serve.

Lithium-ion batteries from Panasonic power the Tesla model S. Source: Reuters/Yuya Shino

On top of that, many of these ECUs run in very harsh environments with wide temperature ranges, require high reliability to protect human life, and must be secure from tampering. That’s a tall engineering order to fill. In the ESL space of EDA tools we have seen the gradual deployment of virtual platforms that are able to model a system of hardware and software together and very early in the design process, in order to both ensure highest quality and verify these complex systems before manufacturing.

To get 100 million lines of code working properly requires modeling at higher levels of abstraction by simulating with Transaction Level Modeling (TLM)and using languages like System C. Automotive engineers must bridge between the thermal/mechanical world and electronics by modeling that can simulate the functional behavior with environment affects on power and thermals. Some of the key concerns for power and thermals in the automotive market are:

• Effective thermal management of automotive electronics assemblies, critical for long reliability.
• Thermal management is the main issue affecting battery performance, lifetime and safety.
• Reducing the energy consumption, especially important to EVs and HEVs where reduced energy consumption improves travel range.
• Weight and size reduction also can be optimized to increase vehicle range.

Docea Power

A company called Docea Power has focused on such a challenge and has a tool flow at the ESL level for both power and thermal modeling along with simulation. Aceplorer is flexible in terms of modeling, because you can model the power behavior of: Digital, Analog, AMS, sensors, batteries, displays, etc.

With a TLM-based approach the decision is yours in terms of model complexity versus accuracy: Un-timed model, approximate timed model, loosely-timed model, or cycle accurate model.

The various team members from software to hardware will choose the appropriate models based on their tasks.

Extending the coverage of the virtual platform with power and thermal virtual models supports:

• Ability to Co-simulate functional with non-functional properties: temperature sensor/battery level reported back to functional simulation
• Extension of the functional coverage to reliability issues
• Extending functional development and validation to power/thermal management software
• Assessment of the impact of dynamic QoS handling on both performance and power
• Exploration of the trade-offs between performance-power-thermal

A pedestrian detection example was created with real code running on a virtual platform in order to evaluate both the thermal and power adaption:

Summary

You don’t have to build a physical prototype and then measure your thermal and power values for automotive and embedded systems, instead consider using an ESL-based approach where you can model and simulate at the earliest stages of design. The benefit of using the Docea Power tools is that you can identify and optimize your system at the ESL level which will give you more confidence that power and thermal goals are met by design.

lang: en_US

Share this post via:

• 
• 
• 
•