|
Abstract
|
|
|---|---|
| Infrared interferometers are used in magnetic fusion devices for measuring the line-integratedelectron density of the plasmas. An FPGA-based processing system is currently being used in the TJ-II infrared interferometer to compute the line-integrated electron density. In high performance Digital Signal Processing (DSP) applications, the computations carried out in the FPGAs are usually performed in fixed point. The floating-point values of the algorithm description must be quantized to their fixed-point counterparts, introducing some deviations with respect to the unquantised case. These deviations are modelled as Round-Off Noise (RON) sources, whose effects are propagated through the different parts of the system. Thus, these quantization operations significantly affect to the maximum attainable system performance. In the TJ-II system, the amount of RON is a limiting factor [1]. Therefore, its analysis and reduction is essential to perform the control operations in real time. The effects of the RON have been traditionally estimated using profiling-based methods. These methods are capable of providing accurate results, but at the expense of extremely long simulation times. Hence, some analytical frameworks have been developed to provide fast and accurate estimates of the quantization effects of the digital implementations. However, existing approaches are not accurate in the general case of non-linear systems with feedback loops. To cope with this limitation, a novel technique based on Modified Affine Arithmetic (MAA) and Polynomial Chaos Expansion (PCE) has been recently developed to provide fast and accurate analysis of the RON [2]. This technique is based on decomposing the random variables into weighted sums of Legendre orthogonal polynomials. Using this method, the contributions of the random signals are propagated through the non-linear system, while the correlations among them are preserved. This paper will provide and discuss a variety of quantization configurations for the TJ-II infrared interferometer using Legendre PCE. The theoretical background of the PCE will be introduced first. Next, our approach will be applied to the algorithms of the TJ-II to provide the set of optimal word-lengths of the variables and the functional units. Finally, the results of the implementation of the algorithms in FPGA devices will be presented. In addition, it will be shown that our approach obtains speedups between 8 and 400 times faster with respect to profiling for several testbenches [2]. | |
|
International
|
Si |
|
Congress
|
IEEE 18th Real Time Conference, RTC'12 |
|
|
960 |
|
Place
|
Berkeley (California, USA) |
|
Reviewers
|
Si |
|
ISBN/ISSN
|
978-1-4673-1082-6 |
|
|
|
|
Start Date
|
11/06/2012 |
|
End Date
|
15/06/2012 |
|
From page
|
1 |
|
To page
|
5 |
|
|
Proceedings of IEEE 18th Real Time Conference |