Since its inception, the TMS board family has become a household name in the engineering world. Similarly, a state-of-the-art discussion on Digital Signal Processing (DSP) is incomplete without MATLAB integration. Why, then, is there no direct interface available between these two giants? This section explores the traditional applications of both MATLAB and the TMS board, and the contributions our group aims to make to DSP analysis through our project.
Background Material
MATLAB is a numerical computing environment and programming language that is used by more than one million people in industry and academia. MathWorks, the supplier of MATLAB, asserts that “MATLAB allows easy matrix manipulation, plotting of user interfaces, and interfacing with programs in other languages.” (MathWorks, 2006) It is a highly numerically-oriented language designed for performing computationally intensive tasks that other traditional languages such as C, C++ or Fortran do not perform time-effectively.
MATLAB's unique ability to be programmed to perform repetitive analyses in a time-effective manner corresponds to the repetitive nature of digital signals. Its computational speed allows it to be ideally used for real-time signal sampling and analysis as well. Unsurprisingly, MATLAB is among the most popular and widely used languages for the analysis of DSP signals today. The capability of rapid prototyping of digital signals offered by MATLAB is shared with certain digital signal processors, such as the TMS board, and hence it is no surprise that these often complement each other in traditional DSP analysis.
Digital signal processors come under two categories: application-specific and general purpose processors. Application-specific chips are built for performing specific functions accurately and efficiently, whereas general-purpose chips perform multiple functions and may compromise efficiency with functionality. (Singh 2-3)
A programmable digital signal processor is both cost-effective and may be programmed for a wide range of applications. Such a microprocessor is built with an optimized architecture to process sampled data at a high rate. Simple operations, such as calculating the sum of a geometric series, can be performed at much higher speeds than an ordinary microprocessor. For more complicated operations, the speed difference becomes more noticeable. (Singh 2-3)
This is what makes the programmable TI TMS board ideal for DSP signal analysis. The board's capability to process sampled data at high speeds makes it an effective resource for the real-time processing of digital signals. The repetitive nature of signal processing is exploited in the TMS board by incorporating parallel operations and data pipelining. These features give this board the reputation of being fast and high in throughput. (Singh 2-3)
What Has Been Done with MATLAB
MATLAB is used in a wide range of applications, including signal and image processing, communications, control design, test and measurement, financial modeling and analysis, and computational biology. Add-on toolboxes extend the MATLAB environment to solve particular classes of problems in these application areas.
What Has Been Done With The TMS Board
As programmable digital signal processors are becoming more powerful in terms of speed and functionality, their applications are continually dominating today's industry. The TMS board, specifically, is used in a wide spectrum of areas, such as speech and image processing, automotive, control, communication, entertainment, instrumentation and medicine. Typical applications include toys, medical instruments, speech synthesis and recognition systems, audio equalizers, echo cancellers, and robotic controllers.
Traditional interfacing between MATLAB and TMS board and what we plan to do
The absence of a direct MATLAB-to-TMS board interface in the market today inhibits the performance of many DSP projects. The traditional method of simulating such an interface involves the manual execution of all intermediate steps. This includes converting MATLAB source code to intermediate language code (such as ANSI C), which in turn is converted to Assembly code. This Assembly code is finally downloaded on to the TMS board. With this approach, cumbersome consequences are abound.
Consider what happens when a MATLAB code must be modified. To reflect this change on the TMS board, all subsequent intermediate languages must be manually modified. Needless to say, this is time-consuming and inefficient. In complicated DSP systems, where such frequent modifications are inevitable, such handling becomes a labouring and impractical process, and a more reasonable structure is required.
The objective of our project is to provide a tangible solution to these drawbacks. We aim to improve the efficiency of DSP algorithmic simulation through the development of a direct interface between MATLAB and the TI TMS board. Specifically, the TMS 320C6713 board will be used to implement this project. This is a commercial DSP device made by Texas Instruments. It is a processor which, while retaining all the features of the basic processor architecture, provides a number of additional features for improved speed and performance.
Apart from efficiency, such a direct interface will also make the real-time simulation of DSP algorithms possible. Hence, a continuous stream of inputs can be collected, analyzed and processed into outputs simultaneously. This will further improve the time efficiency of DSP analysis, as a stream of data can be analyzed and processed in real-time.
