Recently, many models using reconfigurable optically pipelined buses have been proposed in the literature. A system with an optically pipelined bus uses optical waveguides, with unidirectional propagation and predictable delays, instead of electrical buses to transfer information among processors. These two properties enable synchronized concurrent access to an optical bus in a pipelined fashion. Combined with the abilities of the bus structure to broadcast and multicast, this architecture suits many communication-intensive applications.

The number of processors involved in the systems considered raises the probability of a fault occurring to significant levels. Researchers have proposed fault tolerant algorithms for many parallel architectures. They have not, however, addressed the issue of fault tolerance for reconfigurable models, and more specifically, for any of the optically pipelined models.

In this seminar a basic understanding of the structure and addressing methods of optically pipelined models will be provided. Fundamental algorithms for one specific optical model, the Linear Array with a Reconfigurable Pipelined Bus System (LARPBS) will be presented. Furthermore, some basic algorithms, such as binary prefix sums, compression, sorting, and permutation routing, that are able to tolerate up to N/2 faults on an N-processor LARPBS will be covered. These results will be extended to algorithms in the areas of image analysis and matrix operations.

A curved-shape description method, curvature-angular transform (CAT), was developed for shape grading of sweet potatoes using computer vision. A three-dimensional feature vector can be extracted for shape recognition. The relationship between a sweet potato and its shape feature vectors was explored for shape extraction. The extracted feature vectors from sweet potatoes graded by human inspectors were used to train a learning vector quantization (LVQ) neural network. The performance of this trained network was compared with grading by human inspectors. The result of its application to sweet potato shape grading using machine vision showed that it has great potential for automated grading of irregular shapes such as sweet potatoes.

Imagine traveling in a car that can take off and fly in the air... This dream is as old as the inventions of automobile and airplane. It appears now that the technology needed to realize the dream has become available.

Those who dream by night in the dusty recesses of their minds wake in the day to find that it was vanity. But the dreamers of the day are dangerous men, for they may act on their dreams with open eyes, to make it possible.--T. E. Lawrence

If you are a day dreamer, please join us for a multimedia presentation by Dr. J. Jim Zhu on the past, present and the future of flying automobiles. You will learn about the engineering and scientific basis for a vertical take off and landing (VTOL) and wingless (lifting body) flying car, and the technological challenges in Aeronautical, Mechanical and Electrical Engineering for the 21st Century personal vehicle. Dr. Zhu will also describe a student research/design project leading to a radio-controlled flying automobile model to be embarked on by the Division of Engineering Research.

The talk will address a significant problem in environmental analysis and other inspection areas: The determination of subsurface composition using non-invasive techniques. Ground Penetrating Radar (GPR) data will be used to illustrate the type of results that can be obtained in very shallow depths such as in pavement analysis. The talk will then focus on the problems of using similar techniques to try and see to depths of about one hundred meters. We will present some very recent results based on the continuous wavelet transform.

The demand for wireless communication systems has exploded throughout the world. The limited bandwidth available for these systems has created problems that all wireless providers are working to solve. Other problems faced include complex multipath propagation, limited battery lifetime, limited cell size, high infrastructure and operating costs, and growing demand for services with high data rates.

In seeking schemes solving these problems researchers have turned their attention to adaptive smart antenna systems which employ antenna arrays coupled with signal processing at the base station. By exploiting the spatial dimension they allow multiple mobile terminals to transmit co-channel signals, thereby increasing capacity and extending cell coverage. To operate appropriately, adaptive smart antenna systems should be capable of estimating the array response vector, which represents the unique propagation pattern between the mobile terminal and the antenna array at the base station. For N samples and an M-element array antenna the singular value decomposition (SVD) method requires O(NM²) time to calculate the sample covariance matrix and O(M³) time for the decomposition. Even though the performance of the SVD method is good in general, the computational cost may not be practical for real time implementations.

A fast algorithm for the estimation of the array response vector which requires O(NM) computations is proposed. Statistical analysis of the array gain using the proposed estimation algorithm is performed for the single mobile terminal case. Experimental measurements were performed to validate the proposed method using the smart antenna testbed. Application of the proposed method is also extended to a multiple mobile terminal case for an uplink adaptive smart antenna system making use of code division multiple access (CDMA). Using the channel parameters given by the International Telecommunication Union (ITU) for the third generation (3G) standardization, indoor and vehicular test environments are considered.

The IEEE Student Chapter at LSU will provide a presentation on their "maze-racing" robot that will compete at the annual region five IEEE conference. The presentation will demonstrate a working model of the robot and explain the hardware and software that was developed for the project.

For the past several years, IEEE region 5 has organized a student robotics contest that is held at the annual region conference. This year's conference will be held in St.\ Louis, Missouri and will feature a competition in which the robots must navigate through an unknown maze and find the exit in the least amount of time. The competition itself will actually consist of two rounds: During the first round, the robot is given up to 20 minutes to wander around the maze until it finally finds the exit. Then, during the second round, the robot will use the information it gathered in the first round and race to the exit as fast as possible. The winning robot will be the one that completes round two in the least amount of time. The robot itself was designed and built completely by members of the IEEE Student Chapter. Because of limitations imposed this year on the robot's physical dimensions, the entire robot was built from the ground up without reusing any components from previous robots--which in turn lead to several interesting innovations in the robot's design.

Multicarrier (MC) systems are being proposed and tested for wireless data transmission in applications such as broadband wireless networking and digital broadcasting of audio and video. Modulation and demodulation use local oscillators at the transmitter and receiver that are not perfectly synchronized. It is necessary to estimate carrier frequency offset at the receiver and compensate. For a free running receiver local oscillator, the MC system performance rapidly deteriorates when the carrier frequency offset between transmitter and receiver is greater than a small fraction of the intercarrier spacing. Therefore, high resolution and low variance estimators are necessary. On the other hand, it is desirable to perform carrier offset estimation without bandwidth and power consuming pilot signals. For power loading to improve spectral efficiency and coding in MC communications, multipath channel information is required. In this talk, novel blind carrier frequency offset and channel estimators for OFDM and MC-CDMA to increase bandwidth and power efficiency are presented. The performance of carrier offset estimator under different scenarios is analyzed the system sensitivity to perturbations is examined. These analytical studies are supported by real experimental data obtained on our software radio testbed. The extension of this work to the analysis of the uplink system performance loss due to carrier offset will be addressed.

A Multiple Bus Network (MBN) connects a set of processors via a set of buses. This mode of connecting processors can be viewed as a generalization of point-to-point topologies, and has many advantages over single bus systems. Two important parameters in an MBN are degree and loading. Degree is the largest number of connections to a processor and loading is the largest number of processors connected to a bus. These parameters are important because they determine cost and implementability. The degree corresponds to the number of input/output ports per processor, and it is desirable to keep it as small as possible. The loading limits the rate at which data can be transmitted on the bus.

This presentation addresses the relationship between the running time, degree, and loading of MBNs that run k-ary tree algorithms (binary tree algorithms are a special case of k-ary tree algorithms with k=2). These algorithms are fundamental, and have a large number of applications. Specifically it is shown that constant loading is not possible in optimal-time k-ary Tree MBNs. An \Omega(n/(k log n/k)) lower bound on the loading for running a kⁿ-input problem optimally is presented. It is also shown that if the degree is increased to k+1 or the algorithm is run sub-optimally then it is possible to achieve constant loading and that present MBNs that match these bounds.

This talk will present a plan for developing a radio controlled model of flying automobile Pegasus I at the LSU Division of Engineering Research. Pegasus I is the first step towards the ultimate goal of a wingless, VTOL aeromobile. It will have delta wings which fold up in driving mode. The vehicle will be driven by an electric motor in driving configuration, with a size about 1/3 that of a typical automobile.  The flying configuration will be developed in two phases. In Phase A, Pegasus I-A, a light weight (9 kg) airframe driven by a pusher propeller will be built. Once the flying quality is assured, Pegasus I-B, a heavier (25-35 kg), more rigid airframe powered by two mini turbojet engines will be developed in Phase B. VTOL capability will be added to future incarnations of the Pegasus to reduce the size of, and to eventually eliminate, the wings.

Vehicle components will be developed mainly as student design/research projects. Innovative research and design topics include, but are not limited to, drive-by-wire and fly-by-wire control and actuation systems, direct-drive wheels, a lifting-body airframe, low-power VTOL technology, and dual-use vehicle components such as suspension, propulsion, vehicle control, and avionics. These will be discussed in more detail in the presentation.

It is anticipated that construction of the Pegasus I-A will be complete by the end of Fall 2000, and flight tests be completed by the end of Spring 2001. It is intended to enter the Pegasus I-A in the 2001 NASA-FAA-AFRL National General Aviation Student Design Competition. Pegasus I-B will be developed in another year, followed by future incarnations of the Pegasus.