Another key requirement for embedded system design is that the software platform should address the various algorithm design views common in real-time embedded design. These design views—sometimes referred to "as models of computation"— include:

* Text-based math
* Continuous time simulation
* State diagrams

Graphical dataflow models
These varied views are needed because different elements of an embedded system are best captured using different representations.  For example, state diagrams are the most common way to represent control functions, while text-based math is a better way to represent complex algorithms.  If a tool supports all of these views, it helps minimize the complexity of translating system requirements into a software design.  Over the last several years, graphical programming has evolved to incorporate all of the models of computation mentioned above to better meet the needs of embedded system designers and their varied skill sets.

Innovating software algorithms
Signal processing and control algorithms deployed onto distributed devices add significant value to systems and end products. Therefore the ability and freedom to design and iterate on algorithms is critical to innovation and optimizing system performance. Domain experts are the engineers and scientists that hold the expertise to develop these algorithms and processes. While they understand the mathematics and semantics of the process, they often don't understand how to deploy these processes on the wide variety of embedded computing devices ranging from floating point architectures such as PowerPCs, x86, ARM, etc…, to fixed-point MPU, DSP, or FPGA architectures. 

Implementing these processes usually requires the expertise of embedded computing (either through learning or by hiring another developer.) This embedded developer who understands embedded computing platforms very well; however, he or she may not have complete understanding of the algorithm. This presents a large communication gap and can not only add large inefficiencies to the design process; it can cause serious flaws in the end system or product.

Graphical programming languages deliver on these communication challenges for certain vertical areas such as motion control or digital filter design.  Considering a design task such as designing digital filters, a domain expert can interactively translate a high-level description of the filter parameters, such as those for the low-pass filter response shown in Figure 2 and automatically convert it to fixed point using LabVIEW. Once converted to fixed point, the tool can run simulations to compare fixed and floating point implementations. The filter designer can iterate on his or her design until he or she is satisfied with the simulation results, at which time he or she will hand the code over to the embedded engineer for implementation.

Conversely, if the embedded developer needs to make changes to the algorithm, he or she can go back into the design, make the needed changes and regenerate new LabVIEW or C code.  Using one platform to make this translation is efficient since both experts are working with the same tool all the way from design to implementation.

<< PREVIOUS NEXT>>

• TSMC dismisses reports on SMIC intervention

• Bright days for high-end microprocessor cards

• Sand 9 goes strong with its MEMS

• Tanner EDA ties up with TowerJazz

• Sale of Altis likely soon