University of Maryland | A. James Clark School of Engineering
Department of Electrical & Computer Engineering
Research Review Day
RESEARCH REVIEW DAY POSTERS
Advanced Information Systems and Computing
A "Time Machine" for the Web
Joseph Jaja, Sangchul Song, Mike Smorul
The Web has become the main publication medium world-wide, covering almost every facet of human activity. In many cases, the Web is the only medium where such information is recorded. However, the Web is an ephemeral medium whose contents are constantly changing and new information is rapidly replacing old information, and hence the critical importance of establishing web archives to capture the information that is deemed important in the long term. In this work, we are developing technologies to search the web as it existed during a specific point or period in the past and to explore the web as it existed in the past. This research is conducted in collaboration with the Internet Archive and the Library of Congress.
Web contents are characterized by their link structures, temporal context, and complex object structure. Over time, archived web contents can present unprecedented opportunities for information and knowledge discovery in linking and fusing the appropriate information spread over several contextual domains. In collaboration with the Internet Archive and the Library of Congress, we have developed a prototype system that supports temporally-anchored full-text search queries and that enables navigation of archived web content as it existed in the past. The system has been tested on web crawls of the 108th and 109th of the US Congress provided to us by the Library of Congress.
Urban 3D Modeling on Google StreetView Datasets
Mahesh Ramachandran, Ashok Veeraraghavan, Rama Chellappa
Structure from motion is gaining renewed interest because of interesting and large scale applications such as 3D urban modeling and terrain estimation from unmanned vehicles. These new applications require fast, robust and scalable structure from motion (SfM) algorithms so that large scale models can be generated. Moreover, the ubiquitous presence of additional sensors such as inertial measurement units and global positioning systems in several application domains motivate the need to develop new algorithms for structure from motion in the context of estimation in the
presence of such additional information. We believe that the
appropriate use of such additional information can help us determine (a) how to fuse these additional measurements appropriately with (SfM) algorithms and (b) to determine what sensors are most useful in the context of specific 3D modeling applications. In this paper, we study the benefits of the availability of a specific form of additional information - the vertical direction (gravity) and the height of the camera both of which can be conveniently measured using inertial sensors, and a monocular video sequence for 3D urban modeling. We show that in the presence of this information, the SfM equations can be rewritten similar to a bilinear form in its unknowns. This allows us to derive a fast, scalable and parallelizable SfM algorithm for large scale applications. We propose a fast iterative estimation procedure to recover the structure of both stationary scenes and moving objects. In case the
additional information available from the sensors is noisy, we provide a refinement procedure that tackles noisy metadata under a consistent structure from motion estimation framework. We characterize the algorithm with respect to its reconstruction accuracy, memory requirements and stability and compare the algorithm to other relevant competing approaches. We demonstrate the advantages of the algorithm over traditional bundle adjustment in its robustness to poor initial solutions and also memory requirements. Finally we present SfM results using our algorithm on several real-world datasets including an aerial video sequence from the VIVID dataset and the Google StreetView research dataset. We believe that further research in SfM in the presence of additional information is needed for tackling large scale real-world 3D modeling applications.
Performance Potential of an Easy-to-Program PRAM-On-Chip Prototype Versus State-of-the-Art Processor
Uzi Vishkin, George Caragea, Alexandros Tzannes
We compare the Paraleap FPGA computer, a 64-processor hardware prototype of the PRAM-driven XMT architecture, with an Intel Core 2 Duo processor and show that Paraleap outperforms the Intel processor by up to 13.89x in terms of cycle counts. The comparison favors the Intel design, since the silicon area of an ASIC implementation of the 64-processor XMT design is the same as that of a single core.
Scalability of Multicore Processors
Meng-Ju Wu, Donald Yeung
Applying Moore's law to Tilera's 64-core tiled processor (TILE64) would mean we can expect to have a 1000-core processor in 6 years! When we migrate to processors with 100s - 1000 cores, the major challenge is to develop, debug, optimize, and execute multi-threaded applications efficiently. Obtaining good performance and power efficiency is the ultimate goal. In this research, we are developing a simulation infrastructure and analysis tools to identify the bottlenecks of fully utilizing future chip multi-processors (CMPs). We are also investigating different techniques to solve these bottlenecks.
Perfect Secrecy and Steiner Tree Packaging
Sirin Nitinawarat, Prakash Narayan, Alexander Barg
1. Introduce a (multiterminal) Pairwise Independent Network (PIN) model and study perfect secret key generation using public communication. In this model, a subset of terminals seeks a perfect secret key with the cooperation from the remaining terminals;
2. Characterize the maximum achievable size per observation of perfect secret key for a PIN model;
3. Propose an efficient algorithm for the terminals to generate a perfect secret key. The algorithm is derived from a maximal Steiner tree packing in a multigraph defining the PIN model;
4. The proposed algorithm achieves the perfect secret key capacity when all the terminals seek a perfect secret key.
Cryptographic Tools for Long Term Data Integrity
Joseph Jaja, Mike Smorul, Sangchul Song
The Audit Control Environment (ACE) is a system designed to continually audit digital collections based on a rigorous cryptographic technique. This results in a system that allows any component to be externally audited by an independent third party. ACE has been shown to be highly scalable and able to monitor collections across a variety of collections.
Optimization of Irregular Computations on Multithreaded GPUs Using CUDA
Zheng Wei, Joseph Jaja
Parallel computing receives more and more attention these days and NIVIDA's CUDA introduces a programming environment for hundreds of parallel processors inside GPU. List ranking contains extremely irregular computations that are hard to parallelize. In this research, a randomized algorithm is developed to solve this problem on NIVIDA Tesla GPU and very good performance is achieved.
Automatic Detection of Statically Schedulable Region in Dynamic Systems
Ruirui Gu, William Plishker, Shuvra Bhattacharyya
Synchronous dataflow (SDF) is capable of powerful analysis of digital signal processing (DSP) applications. However, its restricted expressibility prevents it from being applied to more complex applications. The CAL language is more expressive, capable of describing dynamic applications, but it cannot directly utilize static scheduling. This work bridges the divide between these two approaches by automatically detecting statically schedulable regions (SSRs) in dynamic CAL applications. By identifying these regions we can apply SDF scheduling techniques to produce more efficient implementations. We demonstrate that by utilizing SSRs detection and static scheduling, this approach is able to significantly improve the throughput of an IDCT application.
Secondwrite: Better Than First
Kapil Anand, Rajeev Barua
The poster is about Secondwrite - a binary rewriter developed by our research group, its applications and advantages
A Probabilistic Characterization of Pose Uncertainty under Self Occlusion
Ming-Yu Liu, Aswin Sankaranarayanan, Rama Chellappa
We study the uncertainty in human body
pose from 2D silhouette observations. Under perspective imaging, the loss of depth information and self-occlusion
of the articulating limbs result in multiple 3D poses mapping to similar 2D silhouettes. This introduces ambiguity in
using the silhouette for inference, particularly in capturing human motion. To characterize this uncertainty, we present
an algorithm which approximates the probability map from a silhouette to its possible 2D poses by sampling efficiently
on a 2D pose space. Further, we show that the probability maps from different camera views can be geometrically fused to obtain the pose distribution in 3D space. Finally, the entropy of the posterior distribution is used as a measure
for quantizing the amount of information encoded by the silhouette. This has applications in perceptual grouping
of silhouettes and in view selection for multi-camera networks. We also show that the pose uncertainty decreases with number of camera views as well as consecutive frame observations from video.
Scalability of Multicore Processors
Meng-Ju Wu, Hameed Badawy, Donald Yeung
Applying Moore's law to Tilera's 64-core tiled processor (TILE64) would mean we can expect to have a 1000-core processor in 6 years! When we migrate to processors with 100s - 1000 cores, the major challenge is to develop, debug, optimize, and execute multi-threaded applications efficiently. Obtaining good performance and power efficiency is the ultimate goal. In this research, we are developing a simulation infrastructure and analysis tools to identify the bottlenecks of fully utilizing future chip multi-processors (CMPs). We are also investigating different techniques to solve these bottlenecks.
↑ Back to Top
Carbon Nanotube Transistor Sensors for Biochemical Sensing
Konrad Aschenbach, R.D. Gomez
Describes the elements of a versatile, cost-effective and portable biochemical sensor system that is developed here at the ECE department.
High Specificity Binding of Lectins to Carbohydrate- Functionalized Fiber Bragg Gratings: A New Model for Biosensing Application
Geunmin Ryu, Mario Dagenais
The functionalization of an etched fiber Bragg Grating was realized using a carbohydrate-siloxane conjugate. No fluorescence labels were used. High specificity of binding for the lectins concanavalin A and peanut agglutinin, respectively, with the cognate ligand was observed. Quasi-monolayer selective binding of the lectins to the fiber sensor was inferred based on a theoretical analysis of the observed changes in the refractive index. Our results open the way to the use of un-labeled carbohydrate based sensors for the study of the human glycome.
The Neural Representation of Auditory Modulations Relevant to Speech
Nick Asendorf, Marisel Villafane-Delgado
Magnetoencephalography (MEG) is a brain imaging technique that non-invasively measures neurally-generated magnetic fields. Earlier MEG studies have focused on the neural responses to amplitude modulated (AM) auditory signals near 40Hz. Speech signals, however, contain a wide range of modulation rates, most of which are well below 40 Hz. Therefore we seek to characterize the modulation transfer function (MTF) of the human brain at AM frequencies much lower than 40Hz. The present study uses MEG to measure neural responses to pure-tone carrier signals amplitude modulated at frequencies exponentially fluctuating between 3Hz and 60Hz. Analysis of the neural MEG data includes noise reduction, time-frequency analysis to characterize the MTF, and a comparison to the neural response to constant AM stimuli. The maximal neural response was evident at low rate modulations, with the shape of the MTF following that of a shallow low-pass filter. The phase of the neural response was linear, consistent with an 80 ms delay. Neural phase responses to upward and downward sweeps differed by ~ radians for AM frequencies 15-35 Hz. An exponential AM chirp gave a successful estimate of the neural power MTF, closely matching that of the response to constant AM stimuli.
An Optical MEMS Sensor for Catechol Detection
Peter Dykstra, Junjie Hao, Reza Ghodssi
The reported device combines patterned waveguides with a microfluidic channel to create a sensor for the harmful phenol, catechol. In order to detect the presence of the chemical, a unique polysaccharide, chitosan, is selectively deposited in the channel. Blue laser light is coupled through the deposited chitosan film. Catechol in solution is flown through the microfluidic channel and past the immobilized chitosan. By-products of catechol oxidation bind to the chitosan film and lower the output light intensity. The work demonstrates the detection of various catechol concentrations.
A Microfluidic Platform for Optical Monitoring of Bacterial Biofilms
Mariana Meyer, Stephan Koev, Reza Ghodssi
This project presents a microfluidic platform for formation and observation of bacterial biofilms. Many types of bacteria communicate with each other via intercellular molecular signaling, and thereby regulate their behavior according to population. At a threshold population, bacteria aggregate and form a pathogenic matrix, or biofilm. We have developed a platform for optical monitoring of Escherichia coli biofilm growth and response in microfluidics. Optical absorbance of the biofilm was measured using an external photodiode to quantify growth under various flow parameters. Device fabrication parameters and experimental results will be presented. Our goal is to develop this platform into an integrated, compact device with parallel throughput for applications in developing antibiotics.
Neural Representation of Complex Modulations in the Auditory Cortex
Nai Ding, Jonathan Simon
Natural sounds such as speech contain multiple levels and multiple types of temporal modulations. Because of nonlinearities of the auditory system, however, the neural response to multiple, simultaneous temporal modulations cannot be predicted from the neural responses to single modulations. Perhaps even more importantly, the observed neural representations of multiple, simultaneous temporal modulations can disambiguate among competing candidate neural mechanisms, which would otherwise be left undetermined if only the neural representations of single modulations were examined. Here we demonstrate the cortical neural representation of an auditory stimulus simultaneously frequency modulated (FM) at a high rate, fFM ? 40 Hz, and amplitude modulated (AM) at a slow rate, fAM < 15 Hz. Magnetoencephalography recordings demonstrate fast FM and slow AM stimulus features evoke two separate but not independent auditory steady state responses (aSSR) at fFM and fAM respectively. The power, rather than phase locking, of the aSSR of both decreases with increasing stimulus fAM . The aSSR at fFM is itself simultaneously amplitude modulated and phase modulated with fundamental frequency fAM , demonstrating that the slow stimulus AM is not only encoded in the neural response at fAM but also encoded in the instantaneous amplitude and phase of the neural response at fFM . Both the amplitude modulation and phase modulation of the aSSR at fFM are most salient for low stimulus fAM but remain observable at the highest tested fAM , 13.8 Hz. The instantaneous amplitude of the aSSR at fFM is successfully predicted by a model containing temporal integration on two time scales, ~25 ms and ~200 ms, followed by a static compression nonlinearity.
Quantifying the Astrocytoma Cell Response to Candidate Pharmaceuticals from F-ACTIN Image Analysis
Chi Cui, Joseph Jaja
The distribution, directionality and motility of the actin fibers control cell shape, affect cell function and are different in cancer versus normal cells. Quantification of actin structural changes is important for further understanding differences between cell types and for elucidation of the effects and dynamics of drug interactions. We propose an image analysis framework to quantify the F-actin organization patterns in response to different pharmaceutical treatments. The main problem solved was to determine which features to quantify and what quantitative measurements to compute when dealing with unannotated confocal microscopy images. The resultant numerical features were very effective for quantitatively profiling the changes in the spatial distribution of F-actin and facilitating the comparison of different drugs. The analysis software was implemented in matlab(TM). The validation for the segmentation and quantification was done respectively by visual inspection and through correlation to expected biological outcomes. Preliminary results showed uniquely significant increases in cortical F-actin to stress fiber ratio for increasing doses of OSW1 and Schweinfurthin A (SA) and a less marked increase for cephalostatin 1 derivative (ceph). This increase was not observed for the actin inhibitors: cytochalasin B (cytoB) and Y-27632 (Y). Future studies will further validate the algorithms, and elucidate the molecular pathways and kinetics underlying the F-actin changes. This is the first study quantifying different structural formations of the same protein in intact cells. Since many anti-cancer drugs target the cytoskeleton, we believe that the quantitative image analysis reported here will have broad applications to understanding the mechanisms of candidate pharmaceuticals.
Computational Framework for Controlled F-action Stained Confocal Microscopy Image Simulation
Chi Cui, Joseph Jaja
The validation of image analysis methods used in automated image cytometry has become an important topic. Image generation from biological experiments cannot guarantee that we obtain a complete set of test cases for validation purposes while the ground truth obtained from manual inspection for these images is often questioned for human bias. Also, working on hundreds of images by hand is impractical. In this paper we propose a computational framework for generating synthesized confocal microscopy images with F-actin structures stained and highlighted by Green Fluorescent Protein (GFP). The presented simulation software was developed in MatlabTM. It was built primarily to validate our F-actin quantification software developed in house. With several user controllable parameters it could be used as a versatile tool for validating other cell
Nicole Nelson, David Sander, Marc Dandin
We describe the design, fabrication and performance
of a handheld fluorometer. The designed system is portable, easy to use and can be fabricated using standard processes and micro-fabrication steps. The fluorometer consists of a differential active pixel sensor along with an integrated optical filter. The sensor is fabricated in a standard 0.5 ?m 2-poly 3-metal CMOS process while the filter is a high rejection chromophore embedded in a polymer film which is cast onto the chip. The chromophores developed have long pass wavelengths of 400 nm and 540 nm. We have experimentally demonstrated the application of the fluorometer in thee bioassays which measure metabolic activity and viability of biological cells, which are useful for cytotoxicity and pathogen detection applications. This miniaturization of costly laboratory equipment is useful for lab-on-chip applications and the handheld fluorometer can serve as a platform for future applications in microfluorometry.
Carbon nanotube transistor sensors for biochemical sensing
Suchit Bhattarai, Konrad Aschenbach, Dr. Romel Gomez
Carbon Nanotube based field effect transistors (CNTFETs) make for versatile sensor systems with crucial applications in detection of nucleic acid hybridization, anti-body binding, proteins, ionic concentration and pH of solutions. This project addressed the applicability of these CNTFETs in detection of varying concentrations of sodium chloride, and pH of solutions. The sensitivity of the sensor chips obtained through preliminary experiments were quantified. This research was performed during summer 2009 in affiliation with the MERIT BIEN undergraduate research program at University of Maryland. The ongoing research attempts to investigate the performance of our sensor chips upon exposure to various other buffer electrolytes.
Biological Cells are Not Linear: They Do Not Demulate RF Signals with Lower Frequency Sidebands - Your Cell Phone is Not Frying Your Brain!
Christopher Davis, Quirino Balzano, Vildana Hodzic, Robert Gammon
It has been hypothesized that cell phones present a health hazard because biological cells can demodulate an RF carrier and thereby expose tissue to low frequency signals. We have shown definitively that this does not occur.
Surface Plasmon Enhanced Fluorescence on Nanostructured Metal Surfaces Christopher Davis, Ehren Hwang, Igor Smolyaninov
Fluorescent materials have found widespread use in applications for biological and chemical sensing as well as detectors and optical sources.
For many of these applications we would like to control where fluorescence occurs as well as increasing the amount of fluorescent signal.
Surface plasmon polaritons generated by using grating coupled excitation can provide strong enhancement of fluorescence in combination with precise position control.
It is well known that fluorescence from particles near a metal surface depends strongly on the balance between fluorescent emission and quenching due to the proximity of the metal.
Our goal is to explore the behavior of position sensitive fluorescence enhancement by grating coupled surface plasmons and examine, in detail, the role that grating parameters play in this effect.
Rapid Optical SAR Measurements
Christopher Davis, Vildana Hodzic, Robert Gammon, Quirino Balzano
Laser beam deflections produced by the RF energy absorbed in a transparent phantom have been used to provide a true calorimetric measurement of Specific Absorption Rates (SAR) produced by the RF coming from cell phones. The method uses multiple diode lasers to probe the heat pattern along several paths. This allows solution of the inverse problem to determine the temperature (and absorbed energy) contours of the heated region. The competing technology uses RF probes that potentially modify the RF fields. Here the laser beams do not interfere at all with the RF fields nor contribute to the heating. The basic sensitivity levels achieved correspond to about 200 nanoradians deflection and allow detection to 0.100 mW/gm.
Laser Beam Propagation through Turbulent Media: A New Technique for Probing Length Scales in Clear Air Turbulence
Joseph Harris, Robert Gammon, Christopher Davis
We are measuring the effects of clear air turbulence in the single scattering regime. These measurements are non perturbative, contrast well with hot wire anemometry, have small size scale resolution, and provide spatial and temporal measurement capability.
Advanced CMOS Imaging
David Sander, Babak Nouri, Pamela Abshire
Image sensors convert optical images into electric signals and are one of the most prevalent sensors in use today due to their unrivaled massively parallel data acquisition, sensitivity and selectivity. CMOS image sensors have several advantages over other imaging sensors including ease of integration with other electronics, integrated signal processing, low cost, and compatibility with lab-on-a-chip (LOC) systems. Our lab has developed specialized CMOS imagers with a focus on biosystems applications such as lab-on-a-chip systems, image plane processors, flight stabilization for unmanned aerial vehicles, and medical diagnostics. These specialized sensors include imagers optimized for low light detection, low noise detection, pattern classification, mismatch compensation, feature extraction, photon counting, high energy radiation detection, as well as others.
CMOS Biosensors for Cell-Based Sensing
Pamela Abshire, Anshu Sarje, Nicole Nelson, Somashekar Prakash
Biological agents as transducers between stimuli and electronic sensors have applications in healthcare, military defense, scientific research, and other arenas. We are developing an integrated bioelectronic and biophotonic interface known as cell clinics, for capturing and characterizing small groups of living biological cells. Each of these cell clinics is an integrated micro-electro-mechanical system (MEMS) which consists of a cell-sized well with an actuated lid and circuitry for sensing, signal-processing, and actuation. Cell clinics provide an opportunity to characterize many individual cells in parallel, in contrast with traditional techniques which characterize average properties of an ensemble of cells. We've demonstrated several technological advances that permit the successful interface of integrated circuits with biological cells. We have designed and tested integrated sensors that directly monitor cell presence and viability via capacitance measurement and optical imaging, cell tracking via imaging, and cellular response to environmental stimuli via fluorescence sensing and extracellular electrical potentials (for electrogenic cells). We've also demonstrated system-level integration for control of MEMS structures using integrated circuits.
Handheld Sample Preparation for Complex Samples
Pamela Abshire, David Sander, Ben Shapiro, Mario Urdaneta
We are developing miniaturized sample preparation technology to be integrated with existing biochemical detection platforms to realize handheld devices. The technology will isolate and concentrate analytes from a complex sample, present it to the detector, and measure the resulting events. This is challenging because complex samples (saliva, blood, urine) contain compounds that interfere with the underlying binding mechanisms. Our handheld system will purify and concentrate analytes in such samples thus allowing high performance, portable detection. The integrated, handheld system will be based on miniaturized optics and on the manipulation of particles and fluid streams within a microfluidic system.
EMG Feedback for Rehabilitation of Stroke Patients
Pamela Abshire, Avi Bardack, Pratik Bhandari, James Doggett
The goal of this project is to design, build and test an interactive videogame for physical rehabilitation of individuals suffering from hemiparesis in their lower limbs following stroke. Game control will be achieved by monitoring muscle contractions using surface electromyographic (EMG) sensors rather than the position of a videogame controller as in commercially available gaming systems. This feature will be used to ensure that the physical rehabilitation activities correctly train and reinforce the muscle actions required for healthy gait. The hypothesis is that interactive training using an EMG-controlled videogame will not only improve gait quality but also increase the patient's interest and motivation in engaging in the repetitive exercise necessary for successful rehabilitation and thereby increase the likelihood that the patients will retain long-term mobility in the years following their stroke. This research has been conceived and carried out by a team of 12 undergraduate students in the Gemstone honors program at the University of Maryland, College Park, with guidance from their mentor Dr. Abshire and collaborators Drs. McCombe Waller and Whitall from the Department of Physical Therapy and Rehabilitation Science at the University of Maryland School of Medicine.
Adaptive Integrated Circuits
Timir Datta, Pamela Abshire
We present an overview of our research in the area of adaptive Integrated circuits. Floating gate transistors are the enabling technology which allow for analog adaptation. Examples are provided of novel applications of floating gate transistors in adaptive integrated circuits.
↑ Back to Top
Communications and Networks
Network Cell Routing Model for Control of Throughput and Delay of Air Traffic
Alex Nguyen, John Baras
Attempts to manage air traffic by either decreasing delay or increasing flow can have the reverse effect on the other. In the National Airspace System (NAS), flights typically depart at their chosen times, and flow management techniques are implemented in the air in an effort to maximize flow, which could lead to delay for individual aircraft. Other areas of the world allow flights to depart only at predetermined slot times knowing that there is a clear unobstructed path, leading to minimal delays but possibly underutilizing airspace. We propose a new approach in leveraging the highway cell transmission model (CTM) to the airspace in the form of a multi-objective optimization that trades between maximizing throughput and minimizing delay. The model is posed as a multi-commodity traffic flow integer program where the constraints are relaxed slightly from the CTM to examine strategies for achieving optimal throughput and delay.
In practice, the model is envisioned to initially run offline to determine a preliminary solution to the initial state of the system. Solutions for subsequent changes in actual state can be determined by running the model online for the incremental state change. Stochastic events such as convective weather clearing times or capacities can be included in the model to better see the benefits and impacts of pre-positioning traffic to take advantage of possible future clearing.
Neighbor Discovery in a Wireless Sensor Network
Jeongho Jeon, Anthony Ephremides
In randomly deployed networks, such as sensor networks, an important problem for each node is to discover its neighbor nodes so that the connectivity amongst nodes can be established. The process of discovery needs to be done in an energy-efficient manner. Thus, the discovery algorithm may terminate without certainty that all neighbors have been discovered. In this paper, we consider this problem by incorporating the physical layer parameters. Specifically, the pilot signals that nodes transmit are successfully decoded if the strength of the received signal relative to the interference is sufficiently high. Thus, each node must extract signal parameter information from an unknown number of received signals. This problem falls naturally in the purview of random set theory (RST) which generalizes standard probability theory by assigning sets, rather than values, to random outcomes. Equivalently, RST can deal with estimating unknown quantities when one of the unknowns is the number of the unknown quantities. We apply RST to a chosen neighbor discovery algorithm and compare its performance to that of the same algorithm under classical matched filter detection techniques. The contributions in the paper are twofold: first, we introduce the realistic effect of physical layer considerations in the evaluation of the performance of logical discovery algorithms; such an introduction is necessary for the accurate assessment of how an algorithm performs. Secondly, given the double uncertainty of the environment (that is, the lack of knowledge of the number of neighbors along with the lack of knowledge of the individual signal parameters), we adopt the viewpoint of RST which is well suited for such environments and demonstrate its advantage relative to classical estimation methods.
Cooperative Multicasting Protocol with Random Linear Network Coding at the Relay
Anthony Fanous, Anthony Ephremides
A relay (R) is used to help the source (S) to forward its traffic to all 'n' destinations. Intuitively, using NC at the relay will result in increasing the service rate at the relay because if a set (S) of destinations received say packet 'A' and set (S') received packet 'B' then if the relay forwards A+ B, then both sets can decode the other packet by successfully receiving A+B.
The proposed protocol highly increases the maximum stable throughput rate of a source multicasting to 'n' destinations.
This protocol is suitable to be used in Ad-Hoc networks where broadcast property is used by relay nodes to help the source node forwarding its traffic at high stable rate.
Delay-minimal Transmission for Average Power Constrained Multi-access Communications
Jing Yang, Sennur Ulukus
Our objective is to minimize the average delay in a two-user multi-access system with average power constraints at transmitters. We formulate the problem as a constrained optimization problem and develop a two-step scheme to find the optimal solution analytically. The optimal policy has a threshold structure: only when the sum of the queue lengths is greater than threshold, both users transmit.
Evolutionary Cooperative Spectrum Sensing Game
Beibei Wang, K.J. Ray Liu, T. Charles Clancy
Cooperative spectrum sensing has been shown to be able to greatly improve the sensing performance in cognitive radio networks. However, if cognitive users belong to different service providers, they tend to contribute less in sensing in order to increase their own throughput. In this paper, we propose an evolutionary game framework to answer the question of ``how to collaborate'' in multiuser de-centralized cooperative spectrum sensing, because evolutionary game theory provides an excellent means to address the strategic uncertainty that a user/player may face by exploring
different actions, adaptively learning during the strategic
interactions, and approaching the best response strategy under changing conditions and environments using replicator dynamics. We derive the behavior dynamics and the evolutionarily stable strategy (ESS) of the secondary users. We then prove that the dynamics converge to the ESS, which renders the possibility of a de-centralized implementation of the proposed sensing game. According to the dynamics, we further develop a distributed learning algorithm so that the secondary users approach the ESS solely based on their own payoff observations. Simulation results show that the average throughput achieved in the proposed cooperative sensing game is higher than the case where secondary users sense the primary user individually without cooperation. The
proposed game is demonstrated to converge to the ESS, and achieve a higher system throughput than the fully cooperative scenario, where all users contribute to sensing in every time slot.
Opportunistic Cooperation for Improving the Stability Region
Beiyu Rong, Anthony Ephremides
We investigate the composite effects of multipacket reception (MPR) and relaying capability in affecting the stability region of a wireless network. With a general MPR channel, a trade-off arises as to whether to activate the relay simultaneously with the source so that both transmissions might be successful, or to let the relay remain silent to overhear the source's transmission and then activate the cooperation. As such, we consider a two-user multiple-access system where the user with a better user-destination channel may act as the relay for the other. An opportunistic cooperation scheme through scheduling the relay's transmission is proposed. Then the optimal scheduling probability for maximizing the stability region is characterized, and the corresponding stability region is derived. We show that the stability region of the opportunistic scheme may be convex under certain channel conditions, and may strictly outer-bound the stability region of the conventional cooperation scheme.
Dependence Balance Bounds for Gaussian Networks with Feedback and Cooperation
Ravi Tandon, Sennur Ulukus
We obtain a new outer bound on the capacity region of the two-user multiple access channel with generalized feedback (MAC-GF). This outer bound are based on the idea of dependence balance which was proposed by Hekstra and Willems. To illustrate the usefulness of our outer bound, we investigate two channel models. We consider the Gaussian MAC with noisy feedback (MAC-NF) and the Gaussian MAC with user cooperation (MAC-UC). We demonstrate that our outer bounds improves upon the cut-set bound for all non-zero values of the feedback/cooperation noise variances for both these channel models.
Window Size Evolution of TCP Coupled with the MAC and PHY Layers
John Baras, George Papageorgiou
In this work the interaction between the AIMD algorithm of TCP and the random access channel is investigated. In particular, we examine the effects of the MAC and the physical layer of the backward channel on the window sizeevolution of TCP. The problem of coupling the windowsize evolution of TCP with a random access channel is addressed using point processes and the theory of martingales .
Energy Constrained Real Time Wireless Multicasting
Nof Abuzainab, Anthony Ephremides
The delivery of real time application traffic
is challenging due to the stringent delay requirements,
especially in the case of energy-limited wireless devices.
In this work, we consider the problem of transmitting
information under energy and delay constraints over erasure
channels. We are interested in finding the maximum
throughput that can be achieved under such constraints in
multicasting a set of packets by a transmitter to multiple
receivers over independent time varying channels in a
single hop network. As a first step, we consider maximizing
the number of packets that can be successfully delivered
under these constraints in an ARQ fashion. Subsequently
we consider also the option of using Random Network
Coding (RNC) as an alternative to ARQ as the former
is known to produce throughput improvement in the
multicasting case. We formulate the problem as Markov
Decision process and obtain the solution under certain
Efficient and Robust Communication Topologies for Distributed Decision Making in Networked Systems
Pedram Hovareshti, John Baras
Distributed decision making in networked systems depends critically on the timely availability of critical fresh information. Performance of networked systems, from the perspective of achieving goals and objectives in a timely and
efficient manner is constrained by their collaboration and communication structures and their interplay with the networked system's dynamics. We describe efficient communication topologies for distributed decision making and relate them to small world graphs and more generally to expander graphs.
In most cases achieving the system objectives requires many agent
to agent communications. A reasonable measure for system robustness to communication topology change is the number of spanning trees in the graph abstraction of the communication
system. We address the problem of network formation with robustness and connectivity constraints. Solutions to this problem have also applications in trust establishment and the relationship of trust to control.
Performance Modeling and Network Design of Hybrid Satellite/Wireless Networks
Ayan Roy-Chowdhury, Vahid Tabatabaee, John Baras
Analytical loss network model for hybrid satellite/wireless networks - distinct loss models in lower (terrestrial) layer and higher (satellite overlay) layer. Fixed point approximation (FPA) approach to analyze the network performance as a function of the network parameters. Proposed method is significantly faster than discrete event simulation. Methodology for hybrid network design through sensitivity analysis of the performance metrics with respect to the design parameters.
Lightweight Certificate-based Source Authentication for Hybrid Networks
Ayan Roy-Chowdhury, John Baras
In this poster, we highlight the salient points of an efficient authentication protocol for group communication in hybrid wireless networks with a satellite overlay. The proposed protocol uses a new class of lightweight, symmetric-key certificates called TESLA certificate. The certificates bind the identities of the senders to the anchor elements of their key chains; messages from the senders are authenticated by MACs computed with keys from the chain. The satellite is used as the Certificate Authority to generate the certificates. The satellite also acts as the proxy for the senders in disclosing the MAC keys to the receivers in the network. Due
to the use of symmetric MAC functions, the proposed protocol is much less expensive in terms of node processing power and energy compared to digital signatures. The use of the satellite as the CA and the proxy allows strong security mechanisms and fast message verification. We also analyze the performance of the protocol in comparison to public key-based digital signatures.
The work described in this poster is the material for the invention disclosure:
"Method for Efficient Source Authentication for Multicast Communications in Hybrid Satellite/Wireless Networks," Ayan Roy-Chowdhury and John S. Baras, United States Patent and Trademark Office Application 61/171216.
The Gelfand-Pinsker Channel: Strong Converse and Upper Bounds for the Reliability Function
Himanshu Tyagi, Prakash Narayan
We consider a Gelfand-Pinsker discrete memoryless
channel (DMC) model and provide a strong converse for its
capacity. The strong converse is then used to obtain an upper
bound on the reliability function. Instrumental in our proofs is a
new technical lemma which provides an upper bound for the rate
of codes with codewords that are conditionally typical over large
message dependent subsets of a typical set of state sequences. This
technical result is a nonstraightforward analog of a known result
for a DMC without states that provides an upper bound on the
rate of a good code with codewords of a fixed type (to be found
in, for instance, the Csiszar-Korner book)
Spectrum Auction for Cognitive Radio Networks
Yongle Wu, Beibei Wang, K.J. Ray Liu
Dynamic spectrum access (DSA), enabled by cognitive radio technologies, has become a promising approach to improve efficiency in spectrum utilization, and the spectrum auction is one important DSA approach, in which secondary users lease some unused bands from primary users. However, spectrum auctions are different from existing auctions studied by economists, because spectrum resources are interference-limited rather than quantity-limited, and it is possible to award one band to multiple secondary users with negligible mutual interference. To accommodate this special feature in wireless communications, we present a novel multi-winner spectrum auction game not existing in auction literature. As secondary users may be selfish in nature and tend to be dishonest in pursuit of higher profits, we develop effective mechanisms to suppress their dishonest/collusive behaviors when secondary users distort their valuations about spectrum resources and interference relationships. Moreover, in order to make the proposed game scalable when the size of problem grows, the semi-definite programming (SDP) relaxation is applied to reduce the complexity significantly. Finally, simulation results are presented to evaluate the proposed auction mechanisms, and demonstrate the complexity reduction as well.
Tract Variables and their Application for Noise-robust Speech Recognition
Vikramjit Mitra, Carol Espy-Wilson, Hosung Nam
This poster analyzes the noise robustness of vocal tract
constriction variable estimation and investigates their role for
noise robust speech recognition. We implement a simple
direct inverse model using a feed-forward artificial neural
network to estimate vocal tract variables (TVs) from the
speech signal. Initially, we train the model on clean synthetic
speech and then test the noise robustness of the model on
noise-corrupted speech. The training corpus was obtained
from the TAsk Dynamics Application model (TADA),
which generated the synthetic speech as well as their
corresponding TVs. Eight different vocal tract constriction
variables consisting of five constriction degree variables (lip
aperture [LA], tongue body [TBCD], tongue tip [TTCD],
velum [VEL], and glottis [GLO]); three constriction location
variables (lip protrusion [LP], tongue tip [TTCL], tongue
body [TBCL]) were considered in this study. We also explore
using a modified phase opponency (MPO)  speech
enhancement technique as the preprocessor for the TV
estimation to observe its effect upon noise robustness.
Estimated TVs were processed by a Kalman smoother to
reduce the estimation noise. Finally the TV estimation module is tested using a naturally-produced speech that is
contaminated with noise at different signal-to-noise ratios.
The estimated TVs from the natural speech corpus are then
used in conjunction with the baseline features to perform
automatic speech recognition (ASR) experiments. Results
show an average 22% and 21% improvement, relative to the
baseline, on ASR performance using the Aurora-2 dataset
with car and subway noise, respectively. The TVs in these
experiments are estimated from the enhanced speech.
Multiuser Rate Allocation Games for Multimedia Communications
Yan Chen, Beibei Wang, K.J. Ray Liu
How to efficiently and fairly allocate data rate among different users is a key problem in the field of multi-user multimedia communication. However, most of the existing optimization-based methods, such as minimizing the weighted sum of the distortions or maximizing the weighted sum of the PSNRs, have their weights heuristically determined. Moreover, those approaches mainly focus on the efficiency issue while there is no notion of fairness. In this work, we address this problem by proposing a game-theoretic framework, in which the utility/payoff function of each user/player is jointly determined by the characteristics of the transmitted video sequence and the allocated bit-rate. We show that a unique Nash equilibrium (NE), which is proportionally fair in terms of both utility and PSNR, can be obtained, according to which the controller can efficiently and fairly allocate the available network bandwidth to users. Moreover, we propose a distributed cheat-proof rate allocation scheme for users to converge to the optimal NE using alternative ascending clock auction. We further show that the traditional optimization-based approach that maximizes the weighted sum of the PSNRs is a special case of the game-theoretic framework with the utility function defined as an exponential function of PSNR. Finally, we demonstrate the efficiency and effectiveness of the proposed method via several experimental results.
Coalition Formation in Autonomic Networks
Tao Jiang, John Baras
Autonomic networks depend on collaboration between
their nodes for all their functionalities. The nodes, even if
modeled as selfish, gain from such collaboration, in the sense
that they can accomplish functionality and performance that
is impossible to achieve without such collaboration. However,
such gains from collaboration do not come for free. There
are costs for such collaboration incurred by each node (e.g.
energy consumption for forwarding other nodes packets). In this poster we use constrained coalitional games (i.e. collaborative dynamic games subject to constraints or costs for collaboration) to investigate several key problems in autonomic networks including: network formation, efficient topologies from
the perspective of high performance at low cost such as small
world graph topologies, and stability of the formed coalitions.
A Systems Engineering Approach to Wireless Network Design
Senni Perumal, Vahid Tabatabaee, John Baras
In this poster, we develop a framework and a tool to analyze and design Mobile Adhoc Networks (MANETs) with predictable performance for a baseline scenario so as to satisfy mission requirements. Various application areas include military battlefield networks, mobile collaborative robotics, communication networks for disaster relief, vehicular networks and wireless sensor networks. We develop fast performance estimation models for various layers of the wireless network stack using network wide fixed point equations which are equations coupled iteratively on the entire network till convergence to obtain a consistent solution. Once we obtain estimates of performance metrics, we perform robust and optimal design of these wireless networks by obtaining sensitivities of these performance metrics like throughput, delay with respect to network paramters (like routing probabilities). These sensitivities are obtained using various methods including Automatic Differentiation or via analytical formulas.
Component Based Design of Link State Routing Protocols
John Baras, Vahid Tabatabaee, Kiran Somasundaram
We present the design of components of link state routing protocols separately . We define performance metrics for components with quantifiable effect on the overall network performance. We develop models and methodologies for components that provide desirable performance.
Trust and Collaboration in Autonomic Networks
Tao Jiang, John Baras
Autonomic networks are decentralized and self-organized. Without global knowledge on the states of autonomic networks, it is difficult to predict behaviors of such networks and thus to conduct proper network management and control. This poster focuses on a specific application in autonomic networks: distributed trust management. Two components are studied: trust credential distribution and trust evaluation. Furthermore, autonomic networks rely on collaboration among users. We show that a well-designed trust management system is helpful to enforce collaboration.
Auction-based Spectrum Sharing Mechanism
Sung Hyun Chun, Richard La
We study the problem of designing a new trading market for dynamic spectrum sharing when there are multiple sellers and multiple buyers. First, we study the interaction among homogeneous buyers of spectrum as a noncooperative game and show the existence of a symmetric mixed-strategy Nash equilibrium (SMSNE). Second, we assume that the sellers employ an optimal mechanism, called the generalized Branco's mechanism, and prove that there exists an incentive for risk neutral sellers of the spectrum to cooperate in order to maximize their expected profits at the SMSNEs of buyers' noncooperative game. Third, we model the interaction among the sellers as a cooperative game and demonstrate that the core of the cooperative game is nonempty. This indicates that there exists a way for the sellers to share the profits in a such manner that no subset of sellers will deviate from cooperating with the remaining sellers. Finally, we demonstrate how we can design a profit sharing scheme that can achieve any payoff vector in the nonempty core of the cooperative game.
Component Based Modeling for Cross-Layer Analysis and Design of Ad-hoc Networks
John Baras, Vahid Tabatabaee, Kaustubh Jain
To jointly design MAC and Routing protocols for wireless ad-hoc networks and analyze their performance.
Performance of MAC and Routing protocols depend significantly on each other.
In most cases, protocol design for routing and MAC is done independently of each other, while analysis is done together, typically using discrete event simulators.
Network simulations take a lot of time and provide little insight into the performance of the protocols.
Develop a component based models of the wireless network that considers cross-layer dependency of performance.
Study the effect of each component on the overall performance of the wireless network, and design each component separately.
Behavior Analysis of Multimedia Social Networks
Wan-Yi Lin, K.J. Ray Liu
This is an introduction to signal and information group's current research on behavior analysis of multimedia social networks. We analyze the impact of human behavior on multimedia systems and by analyzing the user behavior dynamics, we derive the optimal cheat-proof and attack-resistant strategies for various types of multimedia social networks.
Random Key Graphs
Osman Yagan, Armand M. Makowski
Random key graphs, also known as uniform random intersection graphs, appear in application areas as diverse as clustering analysis, collaborative filtering in recommender systems and key distribution in wireless sensor networks. This poster presents some recent results concerning the structure of random key graphs and discusses several possible application areas. Highlights include: (i) A zero-one law for graph connectivity (and its critical scaling) as the number of nodes becomes unboundedly large; (ii) A double exponential result for the probability of graph connectivity; and (iii) Clustering coefficients and the "small world" property of random key graphs.
Channel Modeling for FSO Communications and Sensor Networking Inside Structures
Christopher Davis, Mohammed Eslami, Navik Agrawal
FSO links in interior spaces can be direct (line-of-sight), involve specular reflection or involve diffuse reflection
Line-of-sight links are most efficient
Received power depends on range, laser beam divergence (optical antenna gain), receiver area (including any collection optics)
Omni-directional FSO links are completely alignment insensitive, but energy inefficient
Directed links can provide alignment stability if configured appropriately
Reflection can contribute to link budget in naturally specular ducts inside the aircraft
Diffuse reflection depends on the geometry between source and receiver
Three principal geometries considered
coupled cavities with connecting apertures
A Physics-Based Approach to Self-Organization in Next Generation Broadband Wireless Networks
Christopher Davis, Jaime Llorca, Stuart Milner
This research involves the modeling and optimization of hybrid (directional + omnidirectional) wireless networks
Hybrid Directional Wireless Networks with PAT, Mobility and Topology Control
Christopher Davis, Jaime Llorca, Stuart Milner
This research involves analysis of hybrid networks, which involve:
A High Capacity Network Backbone with Directional Transmission [Free Space Optical and RF (up to Gb/s)] with Beyond Line of Sight Communication and is Dynamically Reconfigurable, plus Low Capacity Terrestrial Networks with Omni-Directional Transmission RF (2-10 Mb/s)
System-level Distribution of Signal Processing Applications in Smart Sensor Networks
Chung-Ching Shen, Shuvra Bhattacharyya, Neil Goldsman
Smart sensor network systems attract interests in applications of environmental monitoring, because they provide not only the conventional capabilities for sensing and data collecting but also the demand and response properties, such as voice signal processing and recognition. Such signal processing applications typically require intensive data processing operations, which are difficult to implement directly in resource-limited sensor nodes. In this work, we present a novel design methodology for modeling and implementing computationally-intensive signal processing applications in a smart sensor network system. This methodology explores efficient dataflow modeling techniques for signal processing applications, investigates workload between computation and communication operations, and provides a new way of balancing computation and communication workloads by distributing such applications among nodes in a smart sensor network system.
Improving Co-Channel Speech Segregation by Sequential Streaming and Separation of Unvoiced Speech
Carol Espy-Wilson, Vijay Mahadevan, Srikanth Vishnubhotla
There are many situations where people are talking at the same time over a single channel. Automatic speech recognition (ASR), speaker identification and enhancement for hearing-impaired listeners requires the separation of the signal into different voice streams. Three scenarios can occur in overlapping speech from two people: (1) both signals are periodic (voiced), (2) both signals are aperiodic (unvoiced) or (3) one signal is voiced and the other is unvoiced. In previous work, we developed a methodology for cases 1 and 2 above. In this poster, we make refinements to address two of the most challenging problems in speech segregation (a) separation of unvoiced speech (case 3) and (b) grouping of segregated speech  within and across voiced regions. The proposed algorithm performs better than all of the unsupervised algorithms in ASR performance. It ranks third in comparison to state of the art supervised approaches.The results show a significant perceptual improvement in terms of the PESQ measure.
↑ Back to Top
Energy and Sustainability
Biofabrication Using the Tobacco Mosaic virus for MEMS Applications
Konstantinos Gerasopoulos, Matthew McCarthy, Reza Ghodssi
This poster reports on the development of novel nanofabrication methodologies for the integration of Tobacco mosaic virus biotemplates into standard micromachining applications. The TMV is a benign plant virus with a very a high surface to volume ratio (300 nm length and 18 nm diameter) which can be modified genetically to create strong bond with metal particles. This allows its use as template nanostructure in applications that require increased surface area, such as energy storage devices and nanosensors. In this work, the robustness of the TMV template in various processing conditions is utilized for the incorporation of the biologically-inspired template synthesis method into conventional semiconductor fabrication. More specifically, selective patterning of TMV onto various substrates through photolithographic techniques as well as assembly of the viral templates onto complex three-dimensional microarchitectures is shown. Finally, the possibility of creating a wide variety of complex nanostructures on the viral surface is demonstrated by means of an Atomic Layer Deposition method. These developed processes bridge the worlds of biotechnology and micro/nanofabrication and enable the future utilization of such inexpensive, high-performance nanomaterials in next-generation electronic devices.
Microball Bearing Tribology
Brendan Hanrahan, Reza Ghodssi
To date, microball bearing platforms have been demonstrated in in a wide array of linear and rotary micro-motors, turbines, actuators, and pumps. Future applications such has high precision actuators or generators place high demands on the performance and reliablility of the novel microball bearing platform. Accordingly, researchers at MSAL have been studying the fundemental tribology (friction, lubrication, and wear) of these systems in order to engineer future high-performance bearing platforms.
Dynamic Thermal Management (DTM) under Soft Thermal Constraints
Bing Shi, Domenic Forte, Ankur Srivastava
Loss of performance and reliability due to unpredictable thermal hotspots has became a major issue. Dynamic thermal management, where the chip operation is controlled during runtime for curtailing thermal emergencies has become an active topic of research. Thermal management can be achieved by controlling processor knobs such as core frequency and supply voltage, scheduling of tasks etc, which in effect, control the power dissipation in different parts of the chip. If we persistently keep core temperature above the constraint then the performance and reliability will certainly suffer. But sporadic violation of thermal constraint will have little or no impact. Moreover, in several scenarios, especially in mission critical systems, allocating as much processing power to a task (even if it leads to reduced chip lifetime) might be more desirable than maintaining acceptable thermal state of the chip while missing critical task deadlines. In this research, we investigate DTM policies (for both single and multi-core processors) that allow the thermal constraint to be violated for a user specified time period under the constraint that the processor be brought back into acceptable thermal state at the end of this time period.
Specifically, we focus on model based methods where we assume that the thermal RC circuit for both single and multicore processor is known.
Curing the Chip Fever: Thermal Sensing and Actuation in Nano-scale Systems
Yufu Zhang, Bing Shi, Ankur Srivastava
Our research addresses the problem of monitoring and managing the thermal stress of today's high performance multi-core processors. We proposed a unified statistical design framework to come up with a thermal sensing and actuation infrastructure which can be embedded into existing high performance processor designs. Our design framework is capable of automatically allocating sensors on the chip, compressing sensors to minimize area/power overhead and combining sensor readings to generate accurate thermal profile for the multi-core processor during its execution. Such an accurate view of the chip thermal state would then be fed back into an actuation unit to dynamically control the chip frequency so that its temperature will not reach the critical point when it will cause detrimental effects to the chip.
Longitudinal Focusing Studies on a Space-Charge Dominated Beam
Brian Beaudoin, Patrick O'Shea, Rami Kishek, IREAP
↑ Back to Top
Robotics and Automation
What Can You Tell from the Shape of a Face?
Pavan Turaga, Rama Chellappa
Understanding and modeling of aging in human faces is an important problem in many real-world applications such as biometrics, authentication and synthesis. In this poster, we consider the role of geometric attributes of faces, as described by a set of landmark points on the face. Towards this end, we show that the space of landmarks can be interpreted as a Grassmann manifold. The warping of an average face to a given face is quantified as a velocity vector that transforms the average to a given face along a smooth geodesic in unit-time. This deformation is then shown to contain important information about the geometry of the face. We show in experiments that exploiting geometric cues in a principled manner provides comparable performance to several systems that utilize both geometric and textural cues . We show results on age estimation using the standard FG-Net dataset and a passport dataset which illustrate the effectiveness of the approach.
Integrated Feedback Control of a Linear Electrostatic Micromotor
Mustafa Beyaz, Nima Ghalichechian, Reza Ghodssi
An optical integrated feedback system involving photodiodes is incorporated into a linear electrostatic micromotor for synchronization and positioning. Integration of photodiodes and through-holes into the micromotor has been achieved through a fabrication process using 11 optical masks. Test results showed the successful position sensing and feedback control providing desired motion without loss of synchronization.
Intelligent Servosystems Lab: Overview
This poster was created to showcase recent projects (both theoretical and experimental) at the Intelligent Servosystems Lab.
Opportunistic Wireless Communications for Distributed Controlled Autonomous Vehicles
Hua Chen, Pedram Hovarshti, John Baras
In this poster, we consider collaborative communication and control problem in a networked system of autonomous vehicles. Our mission is to develop a networked group of small vehicles and sensors operating in dynamic, resource-constrained, adversarial environments. We optimize the trajectories of the autonomous vehicles by solving a distributed control problem, and we study the performance of wireless links between communicating vehicles.
Decentralized Control of Communicating Agents
John Turner, Ashia Wilson
Existing distributed algorithms - for the control and coordination of multi-vehicle systems - assume that each agent possesses unlimited sensing and mobility capabilities. In contrast to prior work, here we focus on the design of distributed coordination algorithms for ant-sized robots that will not possess the sensing and movement capabilities of the existing full-scale ones. Our project also involves the creation of a modular simulation environment, which will be useful as a design aid since it can be used to test the functionality of new algorithms. We are coordinating with other groups at UMD who are working on the fabrication of ant-sized robots, and will assist them in determining what kinds of sensing and communication will be necessary to achieve coordinated movement.
Energy-Based Control of a Flexible Beam
Biswadip Dey, Ravi Banavar
We focus our attention on asymptotically stabilizing a vertically upright flexible beam fixed on a moving cart. The control objective is to attenuate the effect of disturbances on the vertically upright profile of the beam. The control action available is the motion of the cart. By regulating this motion, we seek to regulate the shape of the beam. We set our problem in the port controlled Hamiltonian framework. The interconnection of the flexible beam to the cart is viewed as a power conserving interconnection of an infinite dimensional system to a finite dimensional system. The energy Casimir method is employed to obtain the controller.
A Neuromorphic Head Direction System
Tarek Massoud, Timothy Horiuchi
In the poster we present a a neuromorphic VLSI system implementing an attractor-based model for the biological head direction cell system.
Advanced Optimization for Model Predictive Rotorcraft Control
Meiyun He, Andre Tits
We apply constraint reduction to a case study
of quadratic-programming-based model-predictive rotorcraft
control in which, indeed, constraints far outnumber decision
variables. A difficulty is that constraint reduction requires the
availability, for each optimization problem (to be solved online),
of an initial strictly feasible point: Indeed, such points
may not be readily available in the model-predictive control
context. We propose to address this difficulty by substituting
a certain auxiliary, ?1-penalized problem, which has the same
solution as the original problem. As a by-product, this technique
lends itself nicely to the use of "warm starts" that speed up
the solution of the optimization problem.
Optics of Stereovision and Ranging
Christopher Davis, Carole Teolis, John Karvounis, Jared Napora
Simultaneous Localization and Mapping (SLAM) is a problem that has been investigated in robotics for years. To map out a room, or find distances to objects by means of lasers, cameras, navigation units, and other sensors, SLAM has been thoroughly explored.
MOG Laser Vibrometer
Christopher Davis, Kyuman Cho, John Rzasa
This research has involved the development of a laser vibrometer using I-Q demodulation techniques for monitoring the nanoscale vibrations of distant objects.
Gesture Recognition using Trajectories on Manifolds
Mohamed Abdelkader, Wael Abd-Almageed, Rama Chellappa
We address the problem of human gesture modeling and recognition from video data using Riemannian Geometry Tools on the shape space of closed curves.
Each gesture is modeled as a temporal sequence of human poses characterized by the contours of the human silhouettes at each frame.
Under a certain representation, the closed curves representing the contour shapes can be modeled as points lying on a special curved shape space C.
Each Gesture is modeled as a temporal trajectory on this shape space
We propose two different approaches to model these trajectories for recognition:
1- Template Based Approach :
Dynamic Time Warping (DTW) is used to align the different trajectories using the geodesic distance on the shape space.
A gesture template is then calculated by averaging the aligned trajectories on that space.
2- Model Based Approach
We cluster the gesture shape points on the shape space.
We build a non-parametric statistical models to capture the variations within each cluster.
We then model each gesture as a Markov model of transition between these clusters.
Markerless Motion Capture Using Multiple Views
Ashish Shrivastava, Aravind Sundaresan, Rama Chellappa
This Poster describes the human markerless motion capture using multiple views. First, multiple views of a person are captured using calibrated cameras and these views are combined to generate voxels (3D pixels) for human body. These voxels are then transformed in Laplacian Eigenspace domain for segmentation. Finally, human body model is fitted onto segmented voxels.
Robotic Software, SLAM, and Racing
Gil Blankenship, John Karvounis, Jared Napora, Mike Stanley
This research concerns the development of software for the control and autonomous navigation of networked robots. The software is organized as a library of objects and functions for the management of robotic sensors and actuators to enable the accomplishment of specific missions. The capabilities are demonstrated by having the robots participate in races against other autonomous robots and robots with human operators. The ASL software also supports applications of networked robots for Simultaneous Location and Mapping (SLAM) functions.
↑ Back to Top
Coded Strobing Camera
Ashok Veeraraghavan, Dikpal Reddy, Rama Chellappa
In nature, we often come across events which take place in a blink of an eye or faster. Such events cannot be seen or visualized by a human eye. To "freeze" such events devices such as photographic flash and strobes have been used. Further, there are high speed video cameras which help us capture some amazing events which are otherwise impossible to see. But, they are costly and demand lot of resources such as huge storage space for the captured images and more importantly lighting the scene. The question we ask is can we make off-the-shelf cameras capture such events. To our amazement, the answer is yes! Atleast, under certain conditions we can capture high-speed events using normal cameras. Using the emerging and hot theory of compressive sensing, we built a prototype camera which can capture high speed quasi-periodic events. By modulating the incoming visual signal at a camera with random patterns, we capture and then computationally recover the high-speed event. Our camera is significantly light efficient and also uses much less storage space.
Sparsity Inspired Selction and Recognition of Iris Images
Jaishanker K Pillai, Vishal Patel, Rama Chellappa
Iris images acquired from a partially cooperating
subject often suffer from blur, occlusion due to eyelids, and
specular reflections. The performance of existing iris recognition
systems degrade significantly on these images. Hence it is
essential to select good images from the incoming iris video
stream, before they are input to the recognition algorithm.
Here we propose a sparsity based algorithm for
selection of good iris images and their subsequent recognition.
Unlike most existing algorithms for iris image selection, our
method can handle segmentation errors and a wider range
of acquisition artifacts common in iris image capture. We
perform selection and recognition in a single step which is
more efficient than devising separate specialized algorithms
for the two. Recognition from partially cooperating users is a
significant step towards deploying iris systems in a wide variety
Compressive Synthetic Aperture Radar Imaging
Vishal Patel, Glenn Easley, Dennis Healy, Rama Chellappa
We introduce a new Synthetic Aperture Radar (SAR) imaging modality which can provide a high resolution map of the spatial distribution of targets and terrain using a significantly reduced number of needed transmitted and/or received electromagnetic waveforms. This new imaging scheme, requires no new hardware components and allows the aperture to be compressed. It also presents many new applications and advantages which include strong resistance to countermesasures and interception, imaging much wider swaths and reduced on-board storage requirements.
A Chemical Sensing Microsystem Utilizing an Adaptive Feedback Circuit
Xiao Zhu Fan, Nathan Siwak, Reza Ghodssi
A chemical sensor utilizing optical MEMS and a novel feedback circuit is presented. This circuit implementation of a hill climbing feedback algorithm is capable of autonomously detecting resonant frequency shifts for a range of MEMS resonators. Eight different cantilevers have been measured for their resonant frequency. Additionally, the circuit has been used to track resonant frequency shifts due to isopropanol (IPA) adsorption on three different chemical sensors. The frequency detection range, measurement resolution, and sensitivity of the system have been evaluated.
Cantilever Sensor with Integrated Optical Readout
Nathan Siwak, Xiao Zhu Fan, Reza Ghodssi
This project explores the monolithic integration of Indium Phosphide (InP) MicroElectroMechanical System (MEMS) resonators with optical sources for the development of resonant chemical sensors. Our approach presents a chemical sensor system which utilizes the optical properties of the semiconductor InP to monolithically integrate a sensitive optical measurement system with a cantilever resonator sensor. The novelty of this approach lies in the waveguide-integrated InP quantum well lasers used in conjunction with MEMS resonators in a single-substrate fabrication scheme. Chemical detection has been performed with these versatile sensors in the measurement of volatile organic vapors (isopropanol, ethanol). Additionally, the integration of a photodiode readout has been performed successfully. Future work for this project will be focused on fully developing this integrated single chip platform, establishing the first single-chip MEMS sensors made with III-V materials, increasing deployability of these resonators and the sensitivity of existing instruments.
Object Detection: on the Role of Context
Raghuraman Gopalan, Rama Chellappa
To study the role of spatial neighborhood information in detecting objects from images/ videos. Such descriptions can perform better than isolated object descriptors under the presence of inconsistencies in the visual scene. We illustrate this for two object detection problems, (i) to detect human faces by looking for a supporting torso, and (ii) to find lane markings using their adjoining road regions.
Modeling Inter-Person Interactions for Video Based Group Activity Recognition Ruonan Li, Rama Chellappa
While video-based activity analysis and recognition has received much attention, existing body of work mostly deals with single object/person case. Coordinated multi-object activities, or group activities, present in a variety of applications such as surveillance, sports, and biological monitoring records, etc., are the main focus of this research. Unlike earlier attempts which model the complex spatial temporal constraints among multiple objects with a parametric Bayesian network, we propose a Discriminative Temporal Interaction Manifold framework as a data-driven strategy to characterize the group motion pattern without employing specific domain knowledge. In particular, we establish probability densities on the DTIM, whose element, the discriminative temporal interaction matrix, compactly describes the coordination and interaction among multiple objects in a group activity. For each class of group activity we learn a multi-modal density function on the DTIM. A Maximum a Posteriori classifier on the manifold is then designed for recognizing new activities.
Recognition of Quantized Still Face Images
Tao Wu, Rama Chellappa
In this research, we investigate the effects of the number of grey levels on PCA, multiple exemplar discriminant analysis (MEDA) and the elastic bunch graph matching (EBGM) FR algorithms. The inputs to these FR algorithms are quantized images modified by distance and Box-Cox transforms. The performances of PCA and MEDA algorithms are at 87.66% for images in FRGC version 1 experiment 1 after they are thresholded and transformed while the EBGM algorithm achieves only 37.5%. For the same source face images verification problem in document understanding applications, the performances of PCA and MEDA are stable when the images were degraded by noise, downsampling or different thresholding parameters.
Hyperspectral Imagery for People Recognition
Hien Nguyen, Rama Chellappa
We propose the use of hyperspectral cameras, which collect and process information from across hundreds of frequencies, to do people detection and recognition. The information-rich hyperspectral images and the uniqueness of human skin optical spectra enable accurate identification of human even with just a few pixels. The method is invariant to pose, occlusion, robust against illumination, and vigilant against false face masks.
Cooperative Detection of Attacks on MANET Routing
Shanshan Zheng, Tao Jiang, John Baras
This is joint work with Dr. Jiang and Dr. Baras on cooperative detection of attacks on MANET routing.
Robust Face Alignment
Huy Tho Ho, Rama Chellappa
In this poster, we discuss about the face alignment problem. Accurate face alignment is essential to many applications including facial feature extraction, expression analysis and most importantly, face recognition. It is a very challenging task due to the variations in pose, illumination, facial expression as well as the blurring of the faces in the input image.
Face Tracking and Recognition with Camera Network
Ming Du, Aswin Sankaranarayanan, Rama Chellappa
Multi-camera networks are becoming increasingly prevalent and are being used over a wide range
of application domains. For the problem of non-cooperative face recognition, having multiple cameras
provides multiple observations at any given instant, thereby alleviating some of the challenges in the
acquisition of a good image of the face. Invariably, observing a person from a reasonable number of
suitably placed cameras, significantly increases the chance of obtaining a frontal or close to frontal
image of the face. However, achieving pose-invariant face recognition requires either registration of the
face to a model or designing features that are invariant to rotation. In this work, we present a novel
rotational-invariant feature that is built on Spherical Harmonic Decomposition of textural maps of faces
over a sphere. We first track the location of the head using a histogram of the face estimated from
multi-view inputs. The presence of multiple calibrated cameras allows us to track the head in world
coordinates. Once the 3D location of the head is obtained, we can back-project intensity images onto
the sphere to obtain a texture map corresponding to the face. We then show that a rotation-invariant
signature of the face can be obtained by computing the Spherical Harmonics of the texture map over the
sphere. The proposed feature exhibits improved class separation properties. Experiments demonstrates
promising results in face recognition over a dataset of 25 individuals.
Gaussian MIMO Multi-receiver Wiretap Channel
Ersen Ekrem, Sennur Ulukus
We establish the secrecy capacity region of the Gaussian MIMO multi-receiver wiretap channel.
Privacy Preserving Multimedia Retrieval
Wunjun Lu, Min Wu
In this poster, we present techniques which can achieve content-based retrieval over encrypted multimedia databases. By jointly applying signal processing, information retrieval, and cryptography, we can achieve comparable retrieval performance in the encrypted domain compared to plaintext retrieval. In the meanwhile, data privacy is protected from adversaries. This work can have applications in secure online multimedia management.
Nitesh Shroff, Pavan Turaga, Rama Chellappa
Summarizing long unconstrained videos is gaining importance in surveillance, web-based video
browsing, and video-archival. Summarizing a video requires one to identify key aspects which contain
the essence of the video. In this paper, we propose an approach that optimizes two criteria that a video
summary should embody. The first criterion, 'Coverage', requires that the summary be able to represent
the original video well. The second criterion, 'Diversity', requires that the elements of the summary
be as distinct from each other as possible. Given a user-specified summary length, we propose a cost
function to measure the quality of a summary. The problem of generating a Precis is then reduced to a
combinatorial optimization problem of minimizing the proposed cost function. We propose an efficient
method to solve the optimization problem. We demonstrate through experiments (on KTH, unconstrained
skating video, a surveillance video and a YouTube home video) that optimizing the proposed criterion
results in meaningful video summaries over a wide range of scenarios. Summaries thus generated are
then evaluated using both quantitative measures and user-studies.
Trust and Distributed Filtering
Ion Matei, John Baras, Tao Jiang
We consider distributed state estimation of a
linear dynamic systems, observed by various sensors, as
a problem in information fusion. We introduce a novel
model of trust, using weights on the graph links and nodes
that represent the sensor network. These weights can represent
several interpretations of trustworthiness in sensor
networks. We describe three algorithms that integrate distributed
Kalman filtering with these trust weights. We consider
two interpretations of these trust weights as information
accuracy and reliability. We show that by appropriate
use of these weights the distributed estimation algorithm
avoids using information from untrusted sensors. Simulation
experiments further demonstrate the behavior of these
Thin Film Growth by Pulsed Laser Deposition
Agis Iliadis, Saeed Esmaili Sardari
Growth of high quality thin films of various materials including metal-oxide semiconductors on various substrates is extremely important in device fabrication. A promising technique for high quality growth is pulsed laser deposition (PLD). In this method, an excimer laser ablates a target inside a vacuum chamber. The evaporated material then travels toward the substrate and precipitates to form the epitaxial layer.
Currently we are working on zinc oxide. It has been grown under various growth conditions and its properties are being investigated.
Technical summery of our research is as the followings:
KrF excimer laser with a wavelength of 248nm is used
We typically grow 200~400nm thin films
We grow in a low temperature regime (150~350oC)
The growth is carried out inside a vacuum chamber
. Once the films are developed, their structural, optical and electrical properties are investigated by various techniques including XRD, PL, and Hall measurements, respectively.
Soma Biswas, Jie Ni, Rama Chellappa
Face recognition is an active area of research in the field of computer vision. It has wide range of applications in law enforcement and surveillance, information security, etc. Face recognition from natural images has to account for the change in facial appearance due to variations in illumination, pose, expression, age, etc. In this work, we propose a novel approach to recover the intrinsic characteristics of a face, namely, the texture and the shape from a single input image which in turn can be used for face recognition.
Markoc-Directed Key Rekeying
Paul Yu, John Baras, Brian Sadler
The replacement of secret keys is a central problem in key management. Typical solutions involve handshaking messages, complex computations, or trusted third parties. We introduce a novel key replacement method that exploits the randomness of Markov models to efficiently provide fresh keys to the users. Unlike other methods, the proposed method removes the need for extra communications or third parties while having comparable computational requirements, perfect forward secrecy, and resistance to known-key attacks.
Physical Layer (RF) Authentication and Experimental Validation: Single and Multi-Carrier
Paul Yu, John Baras, Brian Sadler
Prior to joining a network, devices must be authenticated. Typical solutions to the authentication problem involve high complexity (e.g. cryptography), high overhead (e.g. Kerberos), or require physical identification (e.g. smart cards). All of these authentication protocols occur in the open, and are subject to attack by adversaries who wish to disrupt the authentication, modify the messages, or impersonate some of the parties.
We propose a class of authentication systems that use properties of the physical layer to provide a system that is stealthy against detection, robust against noise, and secure against attack. The information necessary for authentication is contained in a tag which is hidden by the receiver noise, path loss, shading, multipath fading, etc. present at the receiver. It is only with the secret shared key that the receiver can detect and validate the authentication.
Automated Transportation Infrastructure Analysis, Surveillance, and First-Responder High Definition Wireless Networked Imaging System
Christopher Davis, Stuart Milner
OVERALL AIM: Improved transportation management, incident detection, security, improved highway utilization: current examples
Automatic multiple vehicle tracking
Automatic vehicle identification: model, color, license plate
Per-lane speed measurements
Origin-Destination tracking based on multiple cameras
"Event" detection: crashes, traffic backups, erratic driving, pedestrians
Algorithms for identifying driver behavior
Digital Image Forgery Detection
Matthew Stamm, K.J. Ray Liu
The widespread availability of photo editing software has made it easy to create visually convincing digital image forgeries. As a result, a need has arisen for digital forensic techniques capable of detecting image manipulation. In this poster, we present a method for detecting image forgeries by identifying contrast enhanced regions within an image. Contrast enhancement is frequently used when creating cut-and-paste type image forgeries to ensure that lighting conditions appear consistent throughout the image. We detect contrast enhancement by searching for the intrinsic statistical fingerprints left behind by contrast enhancement mappings. Once the use of contrast enhancement has been identified, we are able to estimate both the contrast enhancement mapping as well as the image's pixel value histogram before manipulation.
Biometric Authentication in Portable Devices
Vladimir Ivanov, John Baras
A biometric authentication that takes place in_ an unsupervised environment (e.g., at home) can be subject to specific types of attacks. Since the biometric information is not secret and not easily changeable, the attacker may obtain the biometric information of the legitimate user and provide it to the biometric sensor. The attacker may also create counterfeited (fake) biometrics. Furthermore, since the device is portable and can be easily stolen, it may be subject to a physical attack. We propose to split the authentication in two steps: the user to a device and the device to a network. For the user-to-device authentication we propose to use fingerprints and a Trusted Platform Module (TPM) that protects the integrity and confidentiality of the data _with hardware support.
We have demonstrated invisibility in two dimensions on a small scale using surface plasmon polaritons.
Multimedia Fingerprinting & Traitor Tracing
Hongmei Gou, Shan He, Ashwin Swaminathan, Avinash L. Varna, and Min Wu
Multimedia piracy is a growing concern for the film industry, resulting in annual losses of several billions of dollars. A promising technique to deter piracy and trace pirates, is "digital fingerprinting", wherein special signals are inserted into each legally distributed copy that can uniquely identify the recipient. When an unauthorized copy is discovered, the embedded signal can be used to determine the person responsible for the leak. Digital Fingerprinting can also be used in security related applications for traitor tracing. This poster summarizes the fingerprinting research by the Media and Security Team (MAST) at UMD.
Multimedia Content Identification
Avinash L. Varna, Wei-Hong Chuang, Ashwin Swaminathan, Min Wu, Elizabeth Do and Cleo M. Schneider
YouTube and other web services alike have revolutionized content sharing and online social networking by providing an easy-to-use platform for users to post and share video. At the same time, content owners have raised serious concerns on unauthorized uploads of copyrighted movies and TV shows to these websites, as witnessed by high-profile lawsuits filed against YouTube and Google. In order to deter copyright violation and more importantly, to help keep online communities alive legally, "content fingerprinting" technologies are deployed to compute a short string of bits to capture unique characteristics of each video and use it determine whether an uploaded video belongs to a set of copyrighted content or not. Content fingerprints are also used by such applications as Shazam on iPhone to use recordings of short audio clips to identify the song and provide information about the artist, the album, and where to buy. This poster summarizes the research on content identification at the University of Maryland.
Privacy-Preserving Multimedia Retrieval
Wenjun Lu, Avinash L. Varna, Ashwin Swaminathan, and Min Wu
We present in this poster about our work on privacy preserving multimedia retrieval. We jointly apply techniques from signal processing, information retrieval, and cryptography to achieve content-based multimedia retrieval over encrypted databases. This work can have important applications in secure online media management.
Thin Film Growth & Device Development
Saeed Esmaili Sardari, Agis Iliadis
High quality thin films of metal-oxide semiconductors like zinc oxide (ZnO) and tin oxide (SnO2) are deposited epitaxially on silicon substrates under various growth conditions by pulsed laser deposition (PLD).
Structural, optical and electrical properties of the films are studied and evaluated by XRD, XPS, PL and Hall measurements. These structures are particularly interesting for sensor applications including gas sensors, mass/force sensors, and optoelectronic detectors. Since the films are deposited directly on Si substrates, standard fabrication procedure can be employed for device development. Contacts of various materials such as gold, platinum, palladium, aluminum, and silver are deposited via thermal evaporation or E-beam evaporation for electrical studies.
We have developed various sensors including biological and gas sensors in the past, and we are developing UV detectors based on ZnO junctions.
Digital Image Forensics
Ashwin Swaminathan, Hongmei Gou, Wei-Hong Chuang, Christine McKay, Min Wu, and K.J. Ray Liu
This poster describes our research on digital image forensics that aims at answering various forensic questions that can be classified into 1) identifying the components within a digital capture device that produces a given image, and 2) discovering the process history a digital image has gone through. For the former, we explain how the algorithms and parameters associated with a certain component within a capture device can be estimated with the given image, and suggest that such estimation can be used to determine the authentic source of the digital image. For the latter, we propose a universal method that can detect if a digital image has been manipulated after capture, and show that different manipulations can actually be distinguished.
Robust Regression Using Sparse Learning
Kaushik Mitra, Ashok Veeraraghavan, Rama Chellappa, CfAR, University of Maryland, College Park, Mitsubishi Electric Research Laboratories
The goal of regression is to estimate the parameter of a model relating two sets of variables, given a dataset. Presence of outliers will make this estimate unreliable and hence the need for robust regression. Algorithms such as LMedS, RANSAC and MSAC are very successful for low-dimensional robust regression, however their combinatorial nature make them practically unusable for high-dimenisional problems. We formulate the robust regression problem by projecting the dependent variable onto the null space of the independent variables. This projection receives significant contributions only from the outliers. The outliers are then identified using (polynomial) sparse learning algorithms. Under certain conditions, these polynomial algorithms are guaranteed to solve the robust regression problem, which is in general a combinatorial problem.
↑ Back to Top
Scene Analysis for Music Understanding
Steven K. Tjoa, K.J. Ray Liu
Music source separation has drawn plenty of attention for its ability to facilitate tasks in music information retrieval. However, source separation is difficult because there does not exist a unique solution; given a musical mixture, there are several valid ways to decompose the mixture into its individual sources.
Researchers have devoted years of effort toward solving the fully automatic blind source separation problem. Although significant progress has been made along the theoretical and algorithmic aspects of source separation, the separation performance achievable by the state of the art is still not adequate for widespread commercial use. Because of a total lack of prior information, blind source separation remains hard to solve.
In our work, we have proposed a novel method for performing source separation of a musical signal given side information in MIDI format of a similar performance. Our method uses nonnegative matrix factorization (NMF) - a popular, convenient, and effective method for decomposing spectrograms. By exploiting the information that MIDI provides, our method is able to separate sources within highly polyphonic musical mixtures. We have also proposed a novel method that imposes additional harmonic constraints upon the musical atoms learned by NMF. When there is significant spectral-temporal overlap among the musical sources, our learning method has better recall and precision than other popular existing matrix factorization methods.
Tunable Filters and Audio Amplifiers
Franklin DeHart, Joseph Gross, Timothy Babich
We have designed electronics to adjust the tone and shape of audio signals. For example, our tunable filters allow you to select a specific range of frequencies within an audio signal and control how prominent they are in the overall sound by either increasing or reducing their amplitude or range.
Design of a Millimeter Wave Imaging Polarimeter
Christopher Davis, Mohammed Eslami, Quirino Balzano
About 95% of what makes up our universe is still unknown. A particular area of focus at NASA Goddard is a future opportunity to obtain more information about gravitational waves created during the big bang. It is believed that measurements of the polarization of the cosmic microwave background will provide a gravitational wave signature that can be used to acquire more knowledge about the universe. However, these measurements are not realizable with current technologies and measurement systems. There are ongoing efforts to make this type of measurement feasible in the next decade.
A Ball and Curved Offset Beam Experiment
Jansen Sheng, Jay Renner, William Levine
The straight beam in the common ball and beam control experiment was replaced by a curved beam mounted away from its center of rotation. The resulting system is much harder to understand, model and control than the ball and straight beam. Nonetheless, the apparatus was constructed, a model was derived using Lagrangian mechanics, and a controller was designed using a linearized model and the LQR. This controller was implemented and successfully stabilized the ball on the curved offset beam.