Real-time Field-programmable Gate Array Implementation Research Articles

FPGA Realization of ANFIS Controller Using a Proposed Digital Design

The design and realization of Adaptive Neuro-Fuzzy Inference System (ANFIS) controller based on Field Programmable Gate Array (FPGA) is presented in this paper. The controller intended to control the temperature of medical oven. A novel design of digital ANFIS is presented here for the implementation process. Different controllers are designed and their results are compared using MATLAB program to show the ANFIS superiority. The designed controllers tested for cell cultures application at 37.5. A reduction is made for the designed digital ANFIS due to the used FPGA limitations. The reduced design minimizes the utilized slices from 366% to 3% and LUTs from 364% to 3%. The reduced design reached an optimum size for this controller to utilize a smallest memory size. A real-time FPGA implementation of the proposed digital ANFIS have been done and verified through Xilinx ISE 14.6 using the VHDL language. The VHDL code for the controller is produced, aggregated and downloaded on the FPGA Spartan 3A/AN FPGA kit. A comparison between the simulation and implementation results is made. The matching between these results proves the effectiveness and robustness of the proposed digital ANFIS and the excellent performance of the FPGA based controller.

High-dynamic wavelength tracking and millimeter-level ranging inter-satellite laser communication link with feedback-homodyne detection.

The Integrated Laser Communication/Ranging System, which uses a coded signal as the ranging information carrier, is of great importance to the next large-capacity inter-satellite information network. In this paper, a system design with a high-sensitivity feedback-homodyne detection scheme and an asynchronous ranging algorithm is demonstrated with real-time field-programmable gate array-implementation (FPGA). The parallel fast Fourier transformation (FFT) estimation is applied to improve the speed and the range of the wavelength drift tracking, which can handle a dynamic wavelength drift up to 2.4pm/s (300MHz/s). Meanwhile, for clock sources with subtle dynamic frequency offset and sufficient stability, the proposed fractional symbol ranging method is proven to achieve millimeter-level measurement accuracy. The designed system is shown to perform well in terms of both laser linewidth tolerance and noise resistance.

Neutron/Gamma discrimination code based on trapezoidal filter

Neutron/gamma discrimination techniques are widely applied in scintillator-based neutron diagnostics for present nuclear fusion tokamak experiments (e.g. JET – Neutron Camera and compact neutron spectrometer) and will also be necessary for neutron diagnostics of up-coming machines (e.g. ITER Radial Neutron Camera). Neutron/gamma discrimination in scintillators relies on the fact that the detectors output pulses have different shapes depending on the impinging particle; several discrimination techniques are described in literature such as the charge-integration, curve-fitting and pattern recognition [1].This paper aims at describing a new technique for neutron/gamma discrimination in scintillators based on trapezoidal filtering, which targets Field Programmable Gate Array (FPGA) implementation due to its recursive nature. Furthermore its capability to restore the baseline of each detected pulse and the fact that the output signals are shorter than the correspondent incoming pulses, points this technique as a promising solution for applications in high count rate conditions. First results coming from the application of a real-time FPGA implementation of the trapezoidal filter to simulated neutron/gamma data including pile-up events and to real scintillator data will be presented.

Field Programmable Gate Array Implementation of Active Control Laws for Multi-mode Vibration Damping

A B S T R A C T This paper investigates the possibility and effectiveness of multi-mode vibration control of a plate through real-time FPGA (Field Programmable Gate Array) implementation. This type of embedded system offers true parallel and high throughput computation abilities. The control object is an aluminum panel, clamped to a Perspex box’s upper side. Two types of control laws are studied. The first belongs to non-model based control. This control law is designed to generate active damping within the designed bandwidth. The second control law is model based H-infinity robust control. A system identification process is needed before the controller comes out. Each of the control laws is implemented on a FPGA target, which is powerful enough to achieve high throughput control loop rates. The experimental control results demonstrate that the non-model based control law has sufficient authority to suppress the interesting modes. The model-based robust control law’s control performance is not so positive compared to the previous method. Therefore, it is not recommended for this application.

Real-time hardware ANN-QFT robust controller for reconfigurable micro-machine tool

This paper shows a reconfigurable micro-machine tool (RmMT) controlled by an artificial neural network based on a robust controller with quantitative feedback theory (QFT). In order to improve the performance of the controller, a field programmable gate array (FPGA) was applied. Since micro-machines present parametric uncertainties under different points of operation, linear controllers cannot deal with those uncertainties. The parametric uncertainties of a micro-machine could be described by a set of linear transfer functions in frequency domain to generate a complete model of the micro-machine; this family of transfer functions can be used for designing a robust controller based on QFT. Although robust control based on QFT is an attractive control methodology for dealing with parametric uncertainties in CNC micro-machines, the real-time FPGA implementation is difficult because robust controllers require a complex discrete representation. In contrast, artificial neural networks (ANNs) work with basic elements (neurons) and run using a parallel topology. Moreover, they are described by simple representation, so the real-time FPGA implementation of ANN controller is a better alternative than the QFT controller. The proposed ANN-QFT controller gives excellent results for controlling the CNC micro-machine tool during the transitory response.

Real-time low-complexity automatic modulation classifier for pulsed radar signals

An automatic modulation classifier (AMC) based on low-complexity signal features and a hierarchical decision tree is presented for pulsed radar applications. This AMC is implemented on a field-programmable gate array (FPGA) platform. The algorithm has been designed to minimize the computational burden to achieve real-time operation. The novelty of the paper consists of the development of a low-computational-complexity radar AMC suitable for real-time FPGA implementation.

Efficient Bayesian Tracking of Multiple Sources of Neural Activity: Algorithms and Real-Time FPGA Implementation

We propose new Bayesian algorithms to automatically track current dipole sources of neural activity in real time. We integrate multiple particle filters to track the dynamic parameters of a known number of dipole sources, resulting in reducing the computational intensity incurred due to the large number of sensors required to observe magnetoencephalography (MEG) or electroencephalography (EEG) measurements. When we also need to estimate the time-varying number of dipole sources, we develop an algorithm based on applying probability hypothesis density filtering (PHDF) for multiple object tracking. The PHDF is implemented using particle filters (PF-PHDF), and it is applied in a closed-loop with MEG/EEG measurements to first estimate the number of sources and then their corresponding amplitude, location and orientation. The PF-PHDF tracking algorithm uses an online, window-based multiple channel decomposition processing approach that reduces the overall processing time and computational complexity. We demonstrate the improved performances of the proposed algorithms by simulating neural activity tracking systems witWe propose new Bayesian algorithms to automatically track current dipole sources of neural activity in real time. We integrate multiple particle filters to track the dynamic parameters of a known number of dipole sources, resulting in reducing the computational intensity incurred due to the large number of sensors required to observe magnetoencephalography (MEG) or electroencephalography (EEG) measurements. When we also need to estimate the time-varying number of dipole sources, we develop an algorithm based on applying probability hypothesis density filtering (PHDF) for multiple object tracking. The PHDF is implemented using particle filters (PF-PHDF), and it is applied in a closed-loop with MEG/EEG measurements to first estimate the number of sources and then their corresponding amplitude, location and orientation. The PF-PHDF tracking algorithm uses an online, window-based multiple channel decomposition processing approach that reduces the overall processing time and computational complexity. We demonstrate the improved performances of the proposed algorithms by simulating neural activity tracking systems with both synthetic and real data. We map the proposed algorithms onto Xilinx Virtex-5 field-programmable gate array (FPGA) platforms and demonstrate real-time tracking performance. For example, our results showed that the PF-PHDF algorithm can process 100 data samples from three dipoles in only 5.1 ms, when 3 dipole sources are present.h both synthetic and real data. We map the proposed algorithms onto Xilinx Virtex-5 field-programmable gate array (FPGA) platforms and demonstrate real-time tracking performance. For example, our results showed that the PF-PHDF algorithm can process 100 data samples from three dipoles in only 5.1 ms, when 3 dipole sources are present.