Engineering has been a part of Rice's curriculum since the university first opened as The Rice Institute in 1912. In those early days Rice offered courses in chemical, civil, mechanical and electrical engineering. Over the years, the engineering program grew, and in 1975 the George R. Brown School of Engineering was established. Today the school is comprised of eight academic departments and includes twelve research institutes and centers. About one fourth of Rice students are engineering majors.

Among the ninety-two engineering faculty are seven members of the National Academy of Engineering. Virtually all undergraduate courses in engineering are taught by Ph.D. faculty. With a small student-to-faculty ratio, students in both programs not only have easy access to professors in the classroom, but work closely with them on research projects. Departments and centers within the school of engineering take advantage of Houston's role as a center for the energy industry, medical research, space exploration, and the city's rapidly growing high-technology sector. Several departments have active industrial affiliates programs, and many research projects are undertaken with local companies. Students benefit from these relationships through collaborative research projects, summer internships, and making contacts for employment before graduation.

The Computer and Information Technology Institute (CITI) at Rice University is a research institution composed of faculty, research scientists, staff, and graduate students dedicated to the advancement of applied interdisciplinary research in the areas of computer and information technology. CITI's goal is to support, foster, and develop research and education across a wide area of computing technologies, computational engineering, and information processing and theory.

CITI's purpose is to facilitate interdisciplinary research across the university's departmental, center, and laboratory boundaries with industrial and government partners, and in collaboration with other universities. Faculty, research scientists, staff, and students associated with CITI are primarily from the Departments of Computer Science, Electrical and Computer Engineering, Computational and Applied Mathematics, and Statistics, although Mathematical Sciences, Biochemistry and Cell Biology, Chemical Engineering, Geophysics, Chemistry, Physics, Space Physics, Civil and Environmental Engineering, and Mechanical Engineering are represented too. There are approximately ninety Ph.D. faculty and research scientists and more than 120 graduate students affiliated with CITI.

James F. Young*

ELEC 201 introduces undergraduates to the principles and practice of engineering design, and to engineering as a profession. Teams of three students design, construct, and program a small autonomous robot to engage in a competition at the end of the course. The course is completely self-contained, assumes no prerequisites, and is intended for all majors, all years. During the course the participants are exposed to issues that confront every practicing engineer: working within constraints, using available technology, design tradeoffs, team dynamics, meeting project specifications, and time constraints.

Patrick O. Murphy, Brandon Essigman, Erik Welsh, Patrick Frantz

Bluetooth is a versatile, inexpensive, low-power wireless technology designed for short-range ad-hoc connections with many potentially useful applications. Two projects are currently underway involving the applicationof Bluetooth in embedded systems. The first project uses Bluetooth to provide real-time traffic information to motorists by sharing information on current and past traffic conditions between moving vehicles. The second project involves the transmission of JPG2000 compressed still images over Bluetooth using a low powered TI DSP.

A.C. Antoulas*, D.C. Sorensen (CAAM Dept.), Serkan Gugercin

Model reduction aims at replacing a system of differential or difference equations of high complexity by one of much lower complexity. In so doing, one tries to preserve certain critical properties of the system (e.g. stability) and approximate well important features (e.g. the system response). During the last two decades, a lot of progress has been made in the theory of this approximation problem. This presentation reviews the foundations of the theory and presents the key results of frequency and time domain approximations (Grammian based balanced truncation and Hankel norm approximation). More recently, the need has arisen to apply these methods to problems of very high complexity; in such cases the resulting computational complexity becomes prohibitively high and different approaches to the problem have to be developed. Techniques that can be applied to large scale systems will also be presented provided the models are sparse or structured (Pade like approximations and Krylov based methods).

Sinan Sinanovic, Michael Lexa, Chris Rozell, Don Johnson*

Systems can be described by their ability to either suppress or transmit information. We investigate a novel approach to quantify information processing using information-theoretic distances, such as the Kullback-Leibler distance. We are currently developing this theoretic background, and examples are given of the analysis of neural communication in the crayfish visual system and the description of digital communication systems.

Michael Wakin, Justin Romberg, Hyeokho Choi, and Richard G. Baraniuk*

Wavelet domain techniques have had great success in image processing. The wavelet coefficients of an image provide a sparse, multiresolution decomposition and can be used to capture salient image structure. Powerful wavelet-based processing tools rely on good statistical models for the coefficients. In image processing, edges convey significant information, but they are also low-dimensional in structure. We examine the spatial constraints imposed by an edge in an image, and relate these to a set of coherent rules (a "grammar") for the wavelet coefficients. We identify a low-dimensional manifold, which describes this coherency, and which can be incorporated into statistical models for the wavelet coefficients. We also describe methods for exploiting this coherency both in complex wavelets for an improved HMT model, and in real wavelets for image compression.

Vinay Ribeiro, Shriram Sarvotham, Mark Coates, Rolf Riedi, Richard Baraniuk*

In networking, probing for `system parameters' have become imperative since direct measurements are either not possible (no cooperation from network) or would lead to forbidding amounts of data. Amongst the probing schemes introduced to infer link speed at a bottleneck router or instantaneous available bandwidth are the packet pairs and bunches or trains, where two or more packets are sent `back to back' onto the network. We present a novel, efficient exponentially spaced probing packet train which operates on an ideal point of the accuracy/sparsity tradeoff, and which provides estimates of the cross-traffic encountered from simple end-to-end observations.

Rui Castro, Mark Coates, Manik Gadhiok, Ryan King, Rob Nowak*, Eric Rombokas, and Yolanda Tsang

Network performance information can be extremely useful, but for most applications it is important that the information be localized. Knowing how portions or individual components of the network are performing is more valuable than generating a global measure of performance. We have developed an innovative and cost-effective process called Network Tomography that produces accurate mappings of network structure and performance. This technology has the potential to impact a number of increasingly important networking tasks including Quality of Service Verification, Maintenance and Provisioning, Content Delivery, and Security. Unfortunately, conducting direct measurement to acquire localized information is an expensive and in many cases impractical task. Because internal routers operate at such high speeds and carry so much traffic, internal measurement demands special purpose hardware devices dedicated to the collection of the traffic statistics. The installation and maintenance of these devices become extremely time-consuming and costly exercises. Whilst measurement throughout a network is infeasible, measurement at the edge of the network is a much more tractable and low-cost task. Techniques that rely only on edge-based measurement allow independent performance monitoring to be performed, because measurement at the edge of the network does not require cooperation from the owner of the network. Network Tomography is the task of inferring the localized performance information and network topology, as demanded by the tasks outlined above, using only edge-based measurements. Without any cooperation from internal routers, our technology determines performance information including the loss rates and delay characteristics of individual links in the network (a link is the connection between two routers). The core idea is the exploitation of the correlation between measurements made at different edge-sites. As proof of concept, we have developed prototype, host-based software that implements our Network Tomography algorithm. The software has been used to conduct experiments on a small network testbed at Rice University, where we can verify that the estimated performance corresponds to the true performance.

Rice Networks Group

Our research goal is to transform best-effort networks such as today's Internet into multi-service networks that support predictable and controllable performance. We address this problem with a combination of algorithm design, theoretical analysis, and prototype implementations. This poster presents a summary of our recent research advances in quality of service addressing (1) scalable services for large scale systems (the Internet), (2) quality-of-service in dynamic and large scale Internet Data Centers, (3) distributed fair scheduling in metro packet rings, (4) edge-based inference of service to validate and monitor the network's QoS capabilities, and (5) opportunistic and fair scheduling on wireless links.

Unoma Ndili, Sridhar Lavu, Rebecca Willett, Erzsebet Merenyi*, Clayton Scott, and Yi Wan

Wavelet bases have nice spatial and frequency localization property, which is common in natural images. Their inherent multiscale nature and fast filter-bank implementation make them suitable for fast processing of images. Using them in statistical frameworks yields new/improved algorithms in image processing such as image segmentation, restoration, enhancement, pattern analysis/synthesis.

Vinay Bharadwaj, Jacob Chakereski, Aditya Dhar, Behnaam Aazhang* and Joseph Cavallaro*

Tarik Muharemovic, Mohammad Jaber Borran, Krishna Kiran Mukavilli, Mahsa Memarzadeh, Behnaam Aazhang* and Joseph Cavallaro*

We propose and analyze several new transmit/receive structures in order to improve the performance of space-time coded systems. For "open loop" systems, where the receiver does not have the channel state information (CSI), we consider the detector-decoder feedback loop in order to improve the system performance. Some prior work has been done in this for maximum likelihood (ML) detectors; however, the complexity of the ML approch is too high. We propose a method for improving the estimates of the MMSE detector, based of the decoder soft outputs. The goal is to improve the performance of the Lucents' VBLAST system. We also adress the problem of designing the multiple antenna transmission schemes using complete and partial channel feedback at the transmitter. Furtehermore, we examine the Space-Time transmission from the broadcast perspective, and we study several transmission beamforming schemes using gaussian information theoretic bounds.

Ahmad Khoshnevis, Nasir Ahmed, Mohammad Ali Khojastepour, Dinesh Rajan, Behnaam Aazhang* and Joseph Cavallaro*

The "Power Control" and "Power Aware Design" projects at Rice include both up-link (from mobile node to base station) and down-link (from base station to mobile node) channels. Here at Rice we work on issues from maximization of battery life, to power aware schedulers, to user cooperation and closed loop power control. I. In optimizing the battery life, the goal is to find ways to increase battery life in wireless devices by optimizing the power consumption at the RF amplifier. II. By introducing a finite delay in design of a power aware scheduler, we reduce the transmission power by waiting for a good channel condition. In our work we show that transmission power can be reduced by utilizing the burstiness of the source. III. By user cooperation, a new kind of diversity, we can reduce the error rate and increase the capacity. This cooperation results in lower symbol error probability without increasing the average user power constraint. IV. In the study of finite rate feedback in down-link power control we show that for a two user system with fading channel and two bits of feedback, up to 15 dB gain can be achieved.

John Brinkmann, Alex Cobb, and Charles Camp

Today's wireless devices are generally designed for use with a single data transmission protocol. A wireless laptop, for example, may operate well within the range of a wireless office network, but will lose its ability to communicate if moved outdoors. Likewise, a cellular telephone may experience poor reception when brought inside the same building. Neither of these devices is capable of dynamically detecting and conforming to differing wireless communication protocols. A more desirable scenario would allow wireless devices to utilize multiple standards, automatically detecting a change in standards and reconfiguring to take advantage of such changes. A mobile laptop could detect that it has left an indoor environment and automatically adjust to utilize the existing outdoor cellular infrastructure. Likewise, a cellular telephone could detect a loss of signal quality and dynamically adjust to utilize an existing indoor network. The RENE mNIC is Rice University's first attempt to construct an interface that detects and adapts to changing wireless standards, and will lay the groundwork for future multi-tiered device research. The first generation mNIC device is currently being assembled, and will support two different transmission standards: a wired Ethernet connection and a USB hub through which different wireless standards will be tested.

Bryan Jones, Chris Steger, Joseph Cavallaro*

This thesis introduces the rapid prototyping methodology, which overcomes barriers in the design and implementation of digital signal processing (DSP) algorithms and systems on embedded hardware platforms, such as a cellular phones. DSP algorithms are described as a set of equations, which are derived from a mathematical description of a problem. The correctness and performance of these algorithms are typically tested by simulating using an equation-friendly software environment, such as Matlab^TM. In contrast, DSP system design is usually described by sketching a block diagram, picturing the interconnection of several blocks. Each block represents a particular algorithm. A software tool which supports the drawing and simulation of block diagrams, such as Simulink, is typically employed. To implement a DSP system, a software tools which operate well with embedded hardware - typically a C compiler and debugger - is used. The tools have little ability to exchange information, leading to barriers between then. A set of rapid prototyping software tools, which improve these communications, have been developed. Using these tools, an advanced cellular phone receiver was designed and implemented. Application of these tools to the design of future DSP systems will shorten development time and improve the correctness of the products.

Vikram Chandrasekhar, Frank Livingston, Mani Vaya, Sridhar Rajagopal, Yuanbin Guo, Behnaam Aazhang* and Joseph Cavallaro*

Code Division Multiple Access (CDMA) schemes are widely accepted for 3rd generation (3G) systems as a bandwidth efficient technique for multi-user wireless communications. Advanced receiver structures for CDMA systems are required to overcome several well-known problems such as the near-far problem and multi-user access interference. The challenges facing the design of an advanced detector for a handset are particularly daunting. The compact size of handsets limits the processing power available for advanced receivers. Furthermore, target data rates specified by the 3G standards impose stringent detection rate requirements upon the handset architecture. Our research at Rice University addresses implementation issues related to the design of advanced detection algorithms in wireless handsets. We have chosen the adaptive blind minimum mean-squared error algorithm (MMSE) for performing the detection. The superior performance of the MMSE algorithm when compared to traditional matched-filtering based detection algorithms, at a relatively inexpensive computational cost, motivated us to choose this algorithm for our study. We have explored design methodologies for chip-serial and chip-parallel architectures of the MMSE algorithm and analyzed the consequent area/time tradeoffs. Analysis of word length effects on the bit error rate demonstrates that a precision of 8 bits is sufficient to achieve near floating-point performance of the algorithm. The proposed architectures exceed data rates for length 31 spreading codes specified by the 3G standards of 384 Kbps (wide area) and 2 Mbps (local area) for a chip-parallel implementation. The implementation was carried out on a Xilinx 800K gate FPGA which provides powerful prototyping capabilities. We are in the process of incorporating our receiver algorithm in the Software radio being developed as part of the RENE project at Rice University. Our ultimate goal is to demonstrate the powerful functionality of the handset detection algorithm in the presence of real world channel fading and multi-path effects and its ability to meet the data rate requirements of current generation multi-media applications.

Martin Leuschen, Ian D. Walker, and Joseph Cavallaro*

This poster displays the application of our recently developed nonlinear analytical redundancy (NLAR) fault detection technique to a two-degree of freedom robot manipulator. NLAR extends the traditional linear AR technique to derive the maximum possible number of fault detection tests into the continuous nonlinear domain. The ability to handle nonlinear systems vastly expands the accuracy and viable applications of the AR technique. The effectiveness of the approach is demonstrated through the two degree of freedom robot example.

Eyal de Lara, Rajnish Kumar, Yuri Dotsenko, Yogesh Chopra, Nilesh Waghela, Dan S. Wallach*, Willy Zwaenepoel*

We want to be mobile, but not at the cost of an hour wait to check an email! We want to use desktop applications, and we need to adapt them to constraints of mobile client. This usually requires source modifications, which is costly. The Puppeteer project implements a novel approach to adaptation we call component-based adaptation which does not need application source modifications. It currently supports adaptation of StarOffice, Internet Explorer, MS-PowerPoint, MS-Word and MS-Outlook. The experiments show that even simple adaptation policies improve the application's response time by almost 80% for opening larger documents. The approach has been extended for doing system-wide adaptation to control multiple applications running concurrently.

Anupam Chanda and Willy Zwaenepoel*

E-commerce sites generate dynamic content using technologies like Java Servlets, PHP scripts, Enterprise Java Beans, etc. In this work we compare the performance of Servlets and PHP for e-commerce sites using the Transaction Processing Performance Councils Web (TPC-W) benchmark. We subject the workload of TPC-W on two e-commerce systems, one implemented in Servlets and the other in PHP and compare the results. Our works show that the performance of the two on TPC-W are practically identical.

Jeremy Pearce and Daniel Mittleman*

We describe the first measurements of the propagation of coherent, single-cycle pulses of terahertz radiation in a scattering medium. By measuring the transmission as a function of the length L of the medium, we extract the scattering mean free path l_s over a broad bandwidth. We observe variations in ls ranging over nearly two orders of magnitude, and covering the entire thin sample regime from L/l_s << 1 to L/l_s ~ 10. We also observe scattering-induced dispersive effects, which can be attributed to the additional path traveled by photons scattered at small angles.

D. C. Larrabee, G. A. Khodaparast, J. Kono, D. S. King, S. J. Chung,and M. B. Santos

By performing two-color pump-probe spectroscopy in a magnetic field, we measured the picosecond (ps) time-resolved cyclotron resonance of transient carriers in InSb quantum wells. After a near-infrared sub-picosecond pump pulse excited electrons in the semiconductor, we observed the change in the absorption of a far-infrared (FIR) probe pulse as a function of magnetic field and time delay. The FIR source was the ps free electron laser at Stanford University. The sample was grown by molecular beam epitaxy and contained 25 periods of 35nm InSb well separated by 50nm of Al0.09In0.91Sb barrier. The sample was mounted on a sapphire window and the GaAs substrate was removed by etching in Br and methanol. These measurements directly monitor the time evolution of the effective mass of relaxing electrons. Our data clearly show that the effective mass decreases as the electrons approach the band edge, consistent with the strong nonparabolicity of the conduction band in InSb. In addition, we observe significant broadening of cyclotron resonance in the early stage of electron relaxation due to increased electron-electron scattering. We compare these results with detailed calculations. Nonparabolicity due to strong coupling between the conduction and valence bands is incorporated using an 8 x 8 Pidgeon-Brown model. The effect of band-gap renormalization on the effective mass will be considered within the Hartree-Fock approximation. The present work will provide new insight into the quantum dynamics of interacting electrons in semiconductors and their transient response to pulsed electromagnetic radiation.

J. Wang, G. A. Khodaparast, B.E. Brinson, and J. Kono

Recently there has been much interest in spin-related phenomena in semiconductors. New device concepts are being sought that employ the spin degree of freedom of electrons in addition to, or in place of, the charge degree of freedom. These "Espintronic" devices will enhance and expand the functionalities of existing devices. InMnAs films and their heterostructures with (Al,Ga)Sb, the first grown III-V dilute magnetic semiconductor (DMS), serve as the prototype for realizing spintronics. The discovery of carrier-induced ferromagnetism in this system has opened up the possibility of integrating information storage and processing capabilities into a single device. Equally importantly, the origin of ferromagnetism is a subject of debate. Various models have been proposed, but no theory can accurately predict Curie temperatures for different DMSs in different density regimes. Unanswered questions include: 1) how the Curie temperature depends on spin-orbit coupling and disorder and 2) whether the carriers responsible for ferromagnetism reside in the valence band or the impurity band. To answer these questions, we are developing a setup to perform time-resolved light-induced magneto-optical Kerr effect (MOKE) experiments. A large density of spin-polarized carriers is created by an ultrafast pump pulse, and a delayed probe pulse monitors the evolution of ferromagnetic order. The MOKE setup employs a polarization balance bridge technique using mercury cadmium telluride detectors to cover wavelengths up to about 10mm, ideal for this narrow-gap system. The details of the setup along with initial results will be presented.

James F. Young

We describe a new program that combines the power of DSP theory and technology with the high-speed, all-optical signal processing capabilities of planar waveguide circuits to provide powerful, adaptable optical signal processing capability. The data path is all-optical and has no speed limitations.

Chu Xiang and James F. Young

The national fiber optic network infrastructure is being challenged by the explosive growth of data traffic and by the need to supply high quality service economically. By far the most important technique for increasing fiber capacity is wavelength division multiplexing (WDM). Unfortunately, WDM, combined with other performance-increasing techniques, results in channel cross talk produced by nonlinear interactions in the fiber. We have developed a novel enhancement to WDM that cancels FWM and other noise, and permits significant improvements in capacity.

J. Qian, J.W. Clark*, K. Lu, F. Ghorbel, A. Bidani and J.B. Zwischenberger

In artificial organ research and development, one of the important problems for researchers is to determine how changes in certain properties of artificial devices affect overall hemodynamics of the animal inquestion. We are particularly interested in the testing ofthe Pericorporeal Artificial Lung (PAL) device, which is being implanted in sheep at UTMB, in Galveston. Simple trial and error approaches prove costly and inefficient. We propose a modeling approach that allows us to mathematically study the cardiovascular system of the sheep and its interaction with the PAL device. Based on published hemodynamic data on sheep, an initial set of parameters has been determined for the model, and it yields very good predictions to typical hemodynamic data from sheep. The model yields biophysically based insights into the nature of the interactions between these two systems, and can be used as a tool for: (a) improving mechanical device design, and (b) guiding experimental design in laboratory.

K. Lu, J.W. Clark*, F. Ghorbel, D. Ware, A. Bidani and J.B. Zwischenberger

We have developed a mathematical model of the human cardiopulmonary system that is able to simulate the normal functions of the cardiovascular and pulmonary systems, as well as their coupled interactions. Included in the model are descriptions of atrial and ventricular mechanics, the hemodynamics of the systemic and pulmonic circulations, baroreflex control of arterial pressure, airway and lung mechanic, gas transport at the lungs and tissue, water exchange at the tissue capillaries, as well as lymphatic flow. With the suitable parameter set, the integrated cardiopulmonary model yielded pressure, volume and flow waveforms that agree well with published data. In addition, the model demonstrated ability to accurately represent heart, lung, and autonomic tone interact during Valsalva Maneuver. It is likely that with further refinement the model could describe various physiological states and help to better understand the biophysics of cardiopulmonary disease.

J. Yang, J.W. Clark*, R.M. Bryan and C. Robertson

This study is concerned with the development of a multiple compartment model of the isolated cerebral artery in rat. The smooth muscle/arterial wall complex is an important component of the circulatory model and serves as a "vasomotor organ", which provides the myogenic mechanism that underlies the phenomenon of the autoregulation of blood flow. We have focused on this myogenic mechanism and have developed a model fof the electrophysiological and contractile characteristics of the single smooth muscle cell of the posterior cerebral artery. This cell model is used to interrelate the topics of arterial wall stress, changes in transmembrane potential, intracellular CA²⁺ concentration and contraction. Moreover, the smooth muscle cell model is embedded in a larger arterial wall model which converts contractile activity into changes in lumen diameter. The complete model consisting of component models of cell, wall, vessel and testing apparatus is used to provide biophysically based explanations of the myogenic mechanisms underlying the autoregulation of cerebral blood flow.