Model-based design replaces paper-passing exercise

Selex Sensors and Airborne Systems has used Simulink to model and implement an advanced nonlinear FPGA-based signal processing and control subsystem for an airborne system project.

Selex Sensors and Airborne Systems (Selex S and AS), a Finmeccanica company, has used Simulink and Simulink Fixed Point from The MathWorks in conjunction with Xilinx System Generator to model and implement an advanced nonlinear FPGA-based signal processing and control subsystem for an airborne system project.

The application of Simulink cut the design time, increased target flexibility, and significantly reduced the need for physical prototypes and proprietary hardware.

Selex S and AS is 75% owned by Finmeccanica and 25% owned by BAE Systems and brings together the former UK based BAE Systems Avionics and Galileo Avionica in Italy.

Systems engineers at Selex S and AS have traditionally relied on a paper-based approach to communicate requirements to hardware engineers.

"Our standard route to an FPGA implementation included a paper-passing exercise between systems engineering and hardware engineers, who then had to interpret specifications before developing a system in VHDL", Alasdair MacLean, Lead Hardware Engineer at Selex S and AS, explains.

"There was a disconnect between the two groups, and we often had to correct misinterpretations".

Reviewing the project MacLean said: "With Simulink, we eliminated the intermediate specification production and interpretation stage, accelerated design iterations, and reduced integration time".

"Design iterations, which would typically take a day or more, were completed in less than one hour - from algorithm to board".

Selex S and AS engineers used Simulink to accurately simulate and verify the system, significantly decreasing the team's dependence on physical prototypes.

"There was a direct correlation between simulation results in Simulink and real-world performance, which eliminated the disconnect between the model requirements and the implementation".

To improve the efficiency of their FPGA design process, Selex S and AS applied MathWorks tools for model-based design with Xilinx System Generator for an airborne image acquisition and stabilisation system.

"With MathWorks tools and Xilinx System Generator, we maintain the same toolset from design to implementation, from the specification stage where the functionality is defined in the Simulink model itself, through to the implementation in the FPGA", says MacLean.

The Selex S and AS engineering team had several options for the airborne system project.

Eddie Power, FPGA and ASIC Design Manager at Selex S and AS, explains: "We had an existing analogue prototype controller, but we felt we needed a nonlinear response, which we could not achieve in an analogue solution".

"A DSP-based software solution required us to use our traditional over-the-wall approach of communicating requirements between systems and hardware engineers".

"Moreover, there were worries that the timing of the DSP would not be fast enough to give the performance bandwidth we needed".

Using tools from MathWorks and Xilinx, Selex S and AS engineers designed and implemented a digital controller for the airborne system using an integrated development environment.

"This approach results in better, faster communication between disciplines, reduces errors caused by misinterpretation of requirements, and cuts development time", notes Power.

Engineers first used Simulink to model the dynamics of the electromechanical hardware, including sensors, actuators, gimbals, and other elements.

They then modelled and simulated the control system in Simulink to verify system performance and data precision.

"It would have been very difficult, if not impossible, to model and then hand code the entire system in VHDL", MacLean notes.

"Simulink gave us a big advantage by enabling us to model the controller and its environment quite accurately".

Selex S and AS translated the model into a fixed-point representation using Simulink Fixed Point.

Once the design was tested, verified, and partitioned in Simulink, engineers then replaced the standard blocks with equivalent blocks from Xilinx System Generator for DSP.

During this conversion process, Selex S and AS engineers ran simulations, ensuring that the functionality and performance had been retained.

Power adds: "Using Simulink and Xilinx System Generator, we raised the level of abstraction and reduced hardware obsolescence issues".

"We can retarget the same design to a different FPGA simply by changing a setting in the model".

Back to top