Janette Frigo

Experience with a Hybrid Processor: K-Means Clustering

We discuss hardware/software co-processing on a hybrid processor for a compute- and data-intensive multispectral imaging algorithm, k-means clustering. The experiments are performed on two models of the Altera Excalibur board, the first using the soft IP core 32-bit NIOS 1.1 RISC processor, and the second with the hard IP core ARM processor. In our experiments, we compare performance of the sequential k-means algorithm with three different accelerated versions. We consider granularity and synchronization issues when mapping an algorithm to a hybrid processor. Our results show that speedup of 11.8X is achieved by migrating computation to the Excalibur ARM hardware/software as compared to software only on a Gigahertz Pentium III. Speedup on the Excalibur NIOS is limited by the communication cost of transferring data from external memory through the processor to the customized circuits. This limitation is overcome on the Excalibur ARM, in which dual-port memories, accessible to both the processor and configurable logic, have the biggest performance impact of all the techniques studied.

Monte Carlo radiative heat transfer simulation on a reconfigurable computer

Recently, the appearance of very large (3 – 10M gate) FPGAs with embedded arithmetic units has opened the door to the possibility of floating point computation on these devices. While previous researchers have described peak performance or kernel matrix operations, there is as yet relatively little experience with mapping an application-specific floating point loop onto FPGAs. In this work, we port a supercomputer application benchmark onto Xilinx Virtex II and Virtex II Pro FPGAs and compare performance with three Pentium IV Xeon microprocessors. Our results show that this application-specific pipeline, with 12 multiply, 10 add/subtract, one divide, and two compare modules of single precision floating point data type, shows speed up of 10.37×. We analyze the trade-offs between hardware and software to characterize the algorithms that will perform well on current and future FPGA architectures.

FPGA implementation of the pixel purity index algorithm

The Pixel Purity Index (PPI) is an algorithm employed in remote sensing for analyzing hyperspectral images. Particularly for low-resolution imagery, a single pixel usually covers several different materials, and its observed spectrum is (to a good approximation) a linear combination of a few pure spectral shapes. The PPI algorithm tries to identify these pure spectra by assigning a pixel purity index to each pixel in the image; the spectra for those pixels with a high index value are candidates for basis elements in the image decomposition. The PPI algorithm is extremely time consuming but is a good candidate for parallel hardware implementation due to its high volume of independent dot-product calculations. This article presents two parallel architectures we have developed and implemented on the Wildforce board. The first one is based on bit-serial arithmetic operators and the second deals with standard operators. Speed-up factors of up to 80 have been measured for these hand-coded architectures. In addition,the second version has been synthesized with the Streams-C compiler. The compiler translates a high level algorithm expressed in a parallel C extension into synthesizable VHDL. This comparison provides an interesting way of estimating the tradeoff between a traditional approach which tailors the design to get optimal performance and a fully automatic approach which aims to generate a correct design in minimal time.

Analog neural network control method proposed for use in a backup satellite control mode

We propose to use an analog neural network controller implemented in hardware, independent of the active control system, for use in a satellite backup control mode. The controller uses coarse sun sensor inputs. The field-of-view of the sensors activate the neural controller, creating an analog dead band with respect to the direction of the sun on each axis. This network controls the orientation of the vehicle toward the sunlight to ensure adequate power for the system. The attitude of the spacecraft is stabilized with respect to the ambient magnetic field on orbit. This paper develops a model of the controller using real-time coarse sun sensor data and a dynamic model of a prototype system based on a satellite system. The simulation results and the feasibility of this control method for use in a satellite backup control mode are discussed.

Radiation detection and situation management by distributed sensor networks

Detection of radioactive materials in an urban environment usually requires large, portal-monitor-style radiation detectors. However, this may not be a practical solution in many transport scenarios. Alternatively, a distributed sensor network (DSN) could complement portal-style detection of radiological materials through the implementation of arrays of low cost, small heterogeneous sensors with the ability to detect the presence of radioactive materials in a moving vehicle over a specific region. In this paper, we report on the use of a heterogeneous, wireless, distributed sensor network for traffic monitoring in a field demonstration. Through wireless communications, the energy spectra from different radiation detectors are combined to improve the detection confidence. In addition, the DSN exploits other sensor technologies and algorithms to provide additional information about the vehicle, such as its speed, location, class (e.g. car, truck), and license plate number. The sensors are in-situ and data is processed in real-time at each node. Relevant information from each node is sent to a base station computer which is used to assess the movement of radioactive materials.

Communications scheduling for concurrent processes on reconfigurable computers

We describe a unified approach to scheduling point-to-point uni-directional communications among concurrent FPGA-based hardware processes. In this model, processes have separate address spaces, and share data through communication. Once a channel is written, it may not be re-written until the receiving process reads the data. Thus if the writer process is ready before the reader has read the previous message, the writer must stall. We present an algorithm to automatically generate synchronized hardware schedules for the parallel processes that communicate, so that hardware stall management is not required. The algorithm requires that the parallel processes conform to certain constraints in program control structures and communications forms. If the processes do not conform to these requirements, hardware-supported stall mechanisms are used. We quantify the impact in area and clock speed between compiler-generated synchronization of process schedules and run-time, hardware-mediated synchronization.

Biologically inspired neural network controller for an infrared tracking system

Many biological system exhibit capable, adaptive behavior with a minimal nervous system such as those found in lower invertebrates. Scientists and engineers are studying biological system because these models may have real-world applications. the analog neural controller, herein, is loosely modeled after minimal biological nervous systems. The system consists of the controller and pair of sensor mounted on an actuator. It is implemented with an electrical oscillator network, two IR sensor and a dc motor, used as an actuator for the system. The system tracks an IR target source. The pointing accuracy of this neural network controller is estimated through experimental measurements and a numerical model of the system.

SATBOT I: Prototype of a biomorphic autonomous spacecraft

Our goal is to produce a prototype of an autonomous robot satellite, SATBOT. This robot differs from conventional robots in that it has three degrees of freedom, uses magnetics to direct the motion, and needs a zero gravity environment. The design integrates the robots structure and a biomorphic (biological morphology) control system to produce a survival- oriented vehicle that adapts to an unknown environment. Biomorphic systems, loosely modeled after biological systems, use simple analog circuitry, are low power, and are microprocessor independent. These analog networks, called nervous networks (Nv), are used to solve real-time controls problems. The Nv approach to problem solving in robotics has produced many surprisingly capable machines which exhibit emergent behavior. The network can be designed to respond to positive or negative inputs from a sensor and produce a desired directed motion. The fluidity and direction of motion is set by the neurons and is inherent to the structure of the device. The robot is designed to orient itself with respect to a local magnetic field; to direct its attitude toward the greatest source of light; and robustly recover from variations in the local magnetic field, power source, or structural stability. This design uses a two neuron network which acts as a push-pull controller for the actuator (air core coil), and two sun sensors (photodiodes) as bias inputs to the neuron. The effect of sensor activation on an attractive or repulsive torque (directional motion) is studied. A discussion of this systems energy and frequency, noise immunity, and some dynamic characteristics is presented.

Advanced processing for high-bandwidth sensor systems

Compute performance and algorithm design are key problems of image processing and scientific computing in general. For example, imaging spectrometers are capable of producing data in hundreds of spectral bands with millions of pixels. These data sets show great promise for remote sensing applications, but require new and computationally intensive processing. The goal of the Deployable Adaptive Processing Systems (DAPS) project at Los Alamos National Laboratory is to develop advanced processing hardware and algorithms for high-bandwidth sensor applications. The project has produced electronics for processing multi- and hyper-spectral sensor data, as well as LIDAR data, while employing processing elements using a variety of technologies. The project team is currently working on reconfigurable computing technology and advanced feature extraction techniques, with an emphasis on their application to image and RF signal processing. This paper presents reconfigurable computing technology and advanced feature extraction algorithm work and their application to multi- and hyperspectral image processing. Related projects on genetic algorithms as applied to image processing will be introduced, as will the collaboration between the DAPS project and the DARPA Adaptive Computing Systems program. Further details are presented in other talks during this conference and in other conferences taking place during this symposium.

Development, implementation, and experimentation of parametric routing protocol for sensor networks

The development of a scalable and reliable routing protocol for sensor networks is traced from a theoretical beginning to positive simulation results to the end of verification experiments in large and heavily loaded networks. Design decisions and explanations as well as implementation hurdles are presented to give a complete picture of protocol development. Additional software and hardware is required to accurately test the performance of our protocol in field experiments. In addition, the developed protocol is tested in TinyOS on Mica2 motes against well-established routing protocols frequently used in sensor networks. Our protocol proves to outperform the standard (MINTRoute) and the trivial (Gossip) in a variety of different scenarios.