MATLAB to Fixed-Point C Code Generation and its Application to Real-Time Heartbeat Detection

MATLAB is a popular very-high-level-language used for visualizing, prototyping and performing design-space exploration of algorithms. But, this flexibility comes at the price of high memory consumption and slow execution times, making it unsuitable for use in an embedded system. The possibility of using Embedded MATLAB (EML), a small subset of the MATLAB language compatible with code generation code, is investigated to generate production level C code from a MATLAB algorithm. The hypothesis of this thesis is that this has the potential of bringing together the best of two worlds; a flexible design interface coupled with a resource-constrained and optimized implementation. In this thesis, a workflow is defined to transform an existing algorithm from an unconstrained MATLAB algorithm to a constrained EML implementation. A new tool is also presented for assisting in automated floating-to-fixed-point conversion of MATLAB algorithms. This conversion will also transfer to the generated C code, thus potentially generating production ready C code. To demonstrate the workflow, a case study on a Real-Time Heartbeat Detection algorithm is presented. It is concluded that although this technology is very promising it still has performance and usability problems that is keeping it from reaching its fullest potential, even if existing workarounds for these problems are included. However, Embedded MATLAB might yet serve its purpose as a bridge between software engineers and scientists, providing a common platform for research and development.