Detail

ASPECTS AND COMPILER OPTIMIZATIONS FOR MATLAB SYSTEM DEVELOPMENT

MATLAB is regarded as a high productivity language contributing to innovative achievements in different areas. However, in the presence of models requiring long simulations, the developer has to code specific portions of the models in less abstract languages (e.g., C). This hampers the high productivity promise, decreases progress, and is also visible when MATLAB specifications are targeted to embedded systems, increasing product development cycles.

This project intends to address aspect-oriented extensions to MATLAB in order to help system modeling and exploration of certain features conceiving system implementation. Aspect mining will be used in order to acquire some aspects relevance from third party MATLAB code.

We hope to identify aspects that will help optimization phases in order to generate performance and memory efficient implementation code.

Code optimizations will be researched in order to inference types, array dimensions and sizes, and memory minimization.

A prototype weaver will be developed in order to evaluate the ideas with complex MATLAB code.