The IPS reseach forum held in Autumn, 2003 provided an overview of current research directions. This page provides a summary of the various presentations for the Forum.
Investigator: Gene T Whipps
Advisor: Randolph L Moses
Sponsor: U.S. Army Research Laboratory
Abstract: The main problem addressed in this work is combined order selection and parameter estimation of coupled harmonic signals in noise. In battlefield applications, many vehicles generate frequency coupled harmonic acoustic and seismic signatures. These features are useful for tracking and classifying battlefield targets. Parameters of frequency coupled harmonics are estimated using nonlinear least-squares (NLS) combined with the Minimum Description Length (MDL).
Summary of current status: whipps-gene-au03.pdf
Authors: Gene T Whipps, Randolph L Moses
Title: A Combined Order Selection and Parameter Estimation Algorithm for
Coupled Harmonics
Conference: MSS Battlefield Acoustic & Seismic Sensing, Magnetic & Electric
Field Sensors, 2003
Status: Published
Authors: Tien Pham, Gene T Whipps
Title: Acoustic Signal Processing Toolbox for Array Processing
Year: 2003
Status: Published
Investigator: Siddharth S Samsi
Advisor: Ashok K Krishnamurthy
Sponsor: This research was made possible through support provided by DoD
HPCMP PET activities through Mississippi State University under the
terms of Agreement No. #GS04T01BFC0060
Abstract: Our research involves the development of a system for deploying Matlab applications over the Web. Using the MATLAB-Java interface available in the recent releases of MATLAB, our system provides the web interface through the use of Java Servlets and custom Java classes. A new MATLAB process is started for every user that logs into the system. All communication between the user and MATLAB process is facilitated by the servlet and Java sockets.
Summary of current status: samsi-siddharth-au00.pdf
Authors: Siddharth S Samsi, Ashok K Krishnamurthy, Stanley C Ahalt, John Nehrbass, Marlon Pierce
Title: A Java based web interface to MATLAB
Conference: HPEC 2003, 2003
Status: Published
Investigator: Parna V Khot
Advisor: Ashok K Krishnamurthy
Sponsor: DoD HPCMP PET
Abstract: The amount and high dimensionality of the data make data mining an important application for high performance computing. Dr. Jeremy Kepner [2002] has developed a system, called MatlabMPI, which allows any Matlab program to exploit multiple processors. My research involves implementation of a parallel clustering algorithm and a parallel classification algorithm using MatlabMPI.
Summary of current status: khot-parna-au03.pdf
Authors: Parna V Khot, Ashok K Krishnamurthy
Title: A Parallel Data Mining Toolbox Using MatlabMPI
Conference: High Performance Embedded Computing Conference, 2003
Status: Published
Investigator: Joshua N Ash
Advisor: Randolph L Moses
Sponsor: Army Research Lab
Abstract: In this paper we present experimental results on propagation, coherence, and time delay estimation (TDE) in an acoustic microphone array. The primary goal is to understand the achievable accuracy of acoustic TDE using low-cost, commercial off the shelf (COTS) speakers and microphones. Based on time delay estimates, resultant sensor network self-localization results are also presented.
Summary of current status: ash-josh-au03.pdf
Authors: Joshua N Ash, Randolph L Moses
Title: Acoustic Sensor Network Self-Localization: Experimental
Results
Conference: Military Sensing Symposia, 2003
Status: Published
Investigator: Maren A. G. Butz
Advisor: Stanley C. Ahalt
Sponsor: Graduate Charities
Abstract: Automated Target Recognition (ATR) is necessary in order to better analyze a situation on the battlefield. ATR allows for faster discovery of unfriendly vehicles which can greatly increase the reaction time of the soldiers in the field. In addition, ATR could decrease the occurrence of friendly fire.
Summary of current status: butz-maren-au03.pdf
Investigator: Kambiz Azarian Yazdi
Advisors: Hesham El Gamal, Philip Schniter
Abstract: In this work, a new cooperative transmission scheme is proposed that enjoys some important advantages over similar protocols. First, it achieves cooperative diversity without relying on orthogonal subspaces (e.g., different spreading codes, non-overlapping time windows or frequency bands), allowing full rate transmission. Second, its performance degrades gracefully as the path loss between cooperating partners increases, never falling below that of non-cooperative protocols.
Summary of current status: azarian-kambiz-au03.pdf
Authors: Kambiz Azarian Yazdi, Hesham El Gamal, Philip Schniter
Title: On the Design of Cooperative Transmission
Schemes
Conference: Allerton Conference on Communication,
Control, and Computing,(Monticello, IL), 2003
Status: Published
Investigator: Praveen Kumar Gopala
Advisor: Hesham El Gamal
Abstract: I have worked on the design of efficient cooperative transmission strategies for the communication of correlated sources over wireless channels in many-to-one sensor networks and the evaluation of their fundamental capacity limits. I am currently investigating the fundamental limits of cooperative protocols for all-to-all and multimodal sensor networks, and the design of cooperative protocols that achieve the same scaling law as those limits.
Summary of current status: gopala-praveen-au00.pdf
Authors: Hesham El Gamal, Praveen Kumar Gopala
Title: On the Scaling Laws of Multi-modal Wireless Sensor Networks
Conference: INFOCOM 2004, 2004
Status: Accepted for publication
Authors: Arul D Murugan, Praveen Kumar Gopala, Hesham El Gamal
Title: Correlated Sources over Wireless Channels: Cooperative Source-Channel Coding
Conference: IPSN 2004, 2004
Status: Under review
Authors: Arul D Murugan, Praveen Kumar Gopala, Hesham El Gamal
Title: Correlated Sources over Wireless Channels: Cooperative Source-Channel Coding
Journal: Journal on Selected Areas of Communication
Year: 2003
Status: Under review
Investigator: Arul D Palanivelu
Advisor: Hesham El Gamal
Sponsor:
Abstract:
Summary of current status: palanivelu-arul-au03.pdf
Authors: Arul Palanivelu, Hesham El Gamal
Title: Correlated Sources over Gaussian Multiple
Access Channels: Blind Cooperative Source Channel
Coding
Conference: Allerton Conference, 2003
Status: Published
Authors: Arul Palanivelu, Praveen Gopala, Hesham El Gamal
Title: Correlated Sources over Wireless Channels :
Cooperative Source-Channel Coding
Journal: Journal on Selected Areas in
Communication
Year: 2003
Status: Under review
Investigator: Solomon F Gibbs
Advisor: Stanley C Ahalt
Sponsor: ARL
Abstract: The support vector machine is a common algorithm for pattern classification. However, many of the most popular implementations are not suitable for parallelization in a distributed-data sense. We discuss current serial techniques and several alternative methods already in literature as well as their potential suitability to the Matlab-MPI system.
Summary of current status: solomon-gibbs-au03.pdf
Investigator: Shengchao Li
Advisor: Philip Schniter
Sponsor:
Abstract: Maximum SNR detectors are proposed for frequency dispersive channel communication, multi-user detection, and MIMO system. Transmit and receive window (pulse shape) designs are proposed for doubly-dispersive channels. My current work involves the pulse shape design for OFDM/Offset QAM system.
Summary of current status: li-shengchao-au03.pdf
Authors: Shengchao Li, Michael P Fitz
Title: On Reduced State Equalizers
Conference: 35'th Asilomar conference on Signals, Systems and
Computers, 2001
Status: Published
Authors: Shengchao Li, Michael P Fitz
Title: Maximum Signal-to-Noise Ratio Breadth-First Multiuser
Detectors
Conference: 2002 Conference on Information Sciences and
Systems, 2002
Status: Published
Authors: Shengchao Li, Michael P Fitz
Title: Finite Delay Maximum SNR Breadth-First Equalizer
Conference: 56th IEEE Vehicular Technology Conference, 2002
Status: Published
Authors: Tai-Ann Chen, Michael P Fitz, Shengchao Li, Michael D Zoltowski
Title: Two Dimensional Space-Time Pilot Symbol Assisted Demodulation for
Frequency Nonselective Rayleigh Fading Channels
Journal: IEEE Transaction on Communications
Year: 2003
Status: Accepted
Authors: Shengchao Li, Michael P Fitz
Title: Maximum SNR Breadth-First Detectors and Equalizers
Journal:
Year:
Status: To be submitted
Investigator: Jing Sun
Advisors: Oscar Y Takeshita, Michael P Fitz
Abstract: Turbo code and LDPC code are two important coding techniques that can work near channel capacity. Interleavers introduce "randomness" in these codes, which are preferrable for good codes. We work on algebraic designs for these interleavers to achieve both better performance and lower memory requirement to store the inteleaver. Permutation polynomial-based interleaver and maximum length linear congruential sequence-based interleaver are proposed for turbo codes and LDPC codes respectively.
Summary of current status: sun-jing-au03.pdf
Authors: Jing Sun, Oscar Y Takeshita
Title: Deterministic Interleavers for Turbo Codes Using Permutation
Polynomials
Journal: IEEE Transactions on Information Theory
Year: 2003
Status: Under review
Authors: Jing Sun, Oscar Y Takeshita
Title: Maximum Length Linear Congruential Sequences-Based Low-Density
Parity-Check Codes
Journal: IEEE Transactions on Communications
Year: 2003
Status: Under review
Investigator: Vivek Rajendran
Advisor: Ashok K Krishnamurthy
Abstract: This work is targeted at developing a signal-processing model to match human performance in the discrimination of dynamic frequency sounds. A computational model that incorporates psychoacoustic knowledge of the human auditory system is used for auditory pre-processing to generate a time frequency (TF) localized, perceptually weighted representation, of the stimuli to be discriminated. A multi-channel detector device is implemented to combine the TF localized information into a single sensitivity measure called d'.The resulting two-stage model is successful in predicting listener performance in complex sound discrimination/formant transition experiments.
Summary of current status: rajendran-vivek-au03.pdf
Authors: Vivek Rajendran, Ashok K Krishnamurthy, Lawrence L Feth
Title: Discrimination of dynamic frequency sounds and a multi-channel
detector model
Conference: The Acoustical Soceity of America, 2003
Status: Published
Authors: Vivek Rajendran, Ashok K Krishnamurthy, Lawrence L Feth, Robert A Fox
Title: A signal processing model to predict the perception of dynamic
speech stimuli
Conference: International Conference on Acoustics,Speech and Signal
Processing, 2003
Status: Under Review
Investigator: Hong Liu
Advisor: Philip Schniter
Abstract: We consider the problem of channel equalization for frequency-selective and time-selective wireless channels. In this scenario, conventional symbol estimation schemes, such as minimum mean-squared error(MMSE) and zero-forcing(ZF) estimation are computationally expensive for large symbol length. We come up with a low complexity equalization algorithm based on iterative MMSE estimation and maximum likelyhood (ML) ratio test.
Summary of current status: liu-hong-au03.pdf
Authors: Philip R Schniter, Hong Liu
Title: Iterative Equalization for Single-Carrier Cyclic-Prefix in Doubly-Dispersive Channels
Conference: Asilomar Conf. on Signals, Systems, and Computers, 2003
Status: Published
Investigator: Arun P Kannu
Advisor: Phil Schniter
Abstract: We consider the problem of channel estimation for Broadband Mobile Wireless systems. The Mean Squared Error(MSE) performance of the pilot based channel estimation is limited by the pilot spacing. So, we need to use previously decoded data in addition to the pilots, to improve the performance. In this scenario, the standard estimators - Kalman and Wiener, become computationally expensive. We come up with a low complexity estimator whose performance is comparable with the standard estimators. Currently, we try to find the optimal training/data transmission scheme and the bounds on the MSE performance.
Summary of current status: kannu-arun-au00.pdf
Investigator: Jonghoon Ryu
Advisor: Oscar Y Takeshita
Sponsor: Graduate Charities
Abstract: In 1997, M. Sudan published new decoding method that can correct errors beyond half the minimum distance. This work was the first algorithm since Berlekamp-Massey algorithm(BMA) in 1960 that performed better in viewpoint of the number of correctable errors. In 1999, V. Guruswami and M. Sudan published even better algorithm which based on the prior work of Sudan. BTW, R. M. Roth showed that Sudan's 1997 paper could be interpreted as the generalized BMA. Our work is focused on if V. Guruswami and M. Sudan's 1999 paper can be interpreted in a similar way as Roth did.
Summary of current status: ryu-jonghoon-au00.pdf
Investigator: Saurabh Sood
Advisor: Ashok Krishnamurthy
Abstract: The accurate identification of characteristic textures is crucial for many applications like classification, indexing, browsing and summarization. Here, graph spectra and graph eigenclusters are proposed as a scalable technique for extracting predominant textures or eigen-textures in a given musical audio. This approach not only makes segmentation more tractable and scalable but also helps in modeling given audio in terms of graphical structure, which is more perceptually revealing.
Summary of current status: sood-saurabh-au03.pdf
Authors: Saurabh Sood, Ashok Krishnamurthy
Title: Extraction of characteristic music textures (eigen-texures) via
graph spectra and eigen clusters
Conference: ICASSP, 2004
Status: Submitted
Investigator: Joonshik Kim
Advisor: Ashock K Krishnamurthy
Sponsor: Graduate Charities
Abstract: The method of extracting the coefficients of filter that models vocal tract is studied. Human speech can be modeled as source-filter model. Glottal pulses correspond to source and vocal tract corresponds to filter. The vocal tract model filter is time varying system. tracking the filter coefficients, and roots - formant of vocal tract - we can apply them in speech recognition and synthesis.
Summary of current status: kim-joonshik-au03.pdf
Investigator: Honglin Li
Advisor: Stanley C Ahalt
Sponsor: DoD High Performance Computing Modernization
Program (HPCMP) Programming Environment and Training (PET).
Abstract: This research proposes a hierarchical structure for video annotation data and investigates the techniques that are necessary to hierarchically extract low-level, mid-level, and high-level video semantic annotation. In particular, we propose to use predictive models in mid-level and a syntactic approach in high-level.
Summary of current status: honglin-li-au03.pdf
Investigator: Adam R Margetts
Advisor: Philip Schniter
Sponsor: Ohio Space Grant Consortium
Abstract: Worldwide, next-generation wireless cellular networks use scrambled multirate CDMA for the physical link. Time-varying echos in the downlink propagation channel destroy orthogonality between the multirate users. We study chip-rate downlink receivers that learn, track, and counter the wireless channel. We derive matched-filter bounds and detect the active spreading-code set.
Summary of current status: margetts-adam-au03.pdf
Authors: Adam R Margetts, Philip Schniter
Title: Chip-Rate Adaptive Two-Stage Receiver for
Scrambled Multirate CDMA Downlink
Conference: Asilomar Conf. on Signals, Systems, and
Computers, 2003
Status: Published
Authors: Adam R Margetts, Philip Schniter
Title: Adaptive Inter-chip Interference Cancellation of Multirate
Scrambled Downlink CDMA
Conference: Proc. Allerton Conf. on Communication, Control,
and Computing, 2003
Status: Published
Authors: Philip Schniter, Adam R Margetts
Title: Adaptive Chip-Rate Equalization of Downlink Multirate
Wideband CDMA
Conference: Asilomar Conf. on Signals, Systems, and
Computers, 2002
Status: Published
Authors: Adam R Margetts, Philip Schniter
Title: Chip-Rate Adaptive Two-Stage Receiver for Scrambled
Multirate CDMA Downlink
Journal: EURASIP Journal on Wireless Communications and
Networking, (Special Issue on Next Generation Cellular CDMA
Systems)
Year: 2003
Status: Submitted
Authors: Adam R Margetts, Philip Schniter
Title: Adaptive Chip-Rate Equalization of Downlink Multirate
Wideband CDMA
Journal: IEEE Transactions on Signal Processing
Year: 2003
Status: Submitted
Investigator: Aditi Kothiyal
Advisor: Oscar Y Takeshita
Abstract: Reed-Solomon (RS) codes are powerful and extensively used error control codes present in every data CD, music CD and DVD. An important and appreciated characteristic of these codes is their beautiful and intricate mathematical structure that lets us to look at them very scientifically. However, an efficient decoding algorithm to match the performance of the optimal maximum likelihood (ML) decoding alogorithm without its complexity, has not been proposed for these codes. The goal of our work is to investigate iterative decoding schemes for RS codes that achieve the performance of ML decoding with a fraction of its complexity.
Summary of current status: kothiyal-aditi-au00.pdf
Investigator: Sibasish Das
Advisor: Phillip Schniter
Abstract: Consider detecting symbols transmitted across a known ISI introducing additive gaussian noise channel. We focus on iterative detection algorithms wherein, an iteration consists of an update of the decision statistic for each symbol utilizing the most recent decision statitic of other symbols in addition to the observation. We derive and show by simulation that the MSE performance of such algorithms approaches the matched filter bound. Presently, we are interested in convergence conditions for these algorithms.
Summary of current status: das-sibasish-au03.pdf
Investigator: Julie Ann Jackson
Advisor: Randolph L Moses
Sponsor: Air Force Research Laboratory; National Science Foundation
Abstract: We found that traditional clutter model distributions (K, Weibull, etc.) were not adequate for lower frequency (VHF) SAR images. We use mixture pdfs to model complex clutter distributions. The Expectation-Maximization (EM) algorithm is used to estimate the proportions and characteristic parameters of the mixture model. The clutter model will be used in evaluating target prediction performance for both single pass and change detection images.
Summary of current status: jackson-julie-au03.pdf
Investigator: Yi Zhao
Advisor: Stanley C Ahalt
Sponsor:
Abstract: Burst packet loss imposes significant quality degradation for streaming applications. Interleaving is an effective method to combat burst error. Most current research focused on how to maximize the interleaving effect in the spatial or spatial-frequency domain. By taking both spatial and temporal interleaving into account, the proposed method focuses on finding a best transmission schedule to maximize to effect of error concealment. Experiment results shows that our proposed method outperforms the existing interleaving approaches.
Summary of current status: zhao-yi-au00.pdf
Authors: Yi Zhao, Honglin Li, Stanley C Ahalt
Title: Video Object Segmentation Using SVMs
Conference: 7th World Multiconference on Systemics, Cybernetics and
Informatics (SCI2003), 2003
Status: Published
Investigator: Heelim Teh
Advisor: Oscar Y Takeshita
Abstract: Channel estimation for Turbo coded TCM signals transmitted through a Rayleigh fading channel using 16QAM is my current research topic. The OSU wireless testbed is designed for vehicle to vehicle communication and is using 220MHz band. So the task is to appy Turbo coding to the current second generation wireless modem.
Summary of current status: teh-heelim-au03.pdf
Investigator: Jingpu J Shi
Advisor: Krishnamurthy K Ashok
Sponsor: Department Of Defense
Abstract: An articulatory speech synthesizer, which includes a voice source model and a vocal tract model, is exploited to produce computer-synthesized speech. The Liljencrants-Fant model is implemented to simulate the voice source. The pitch-synchronous acoustic-to-articulatory transformation, aiming to obtain the shape of the vocal tract from acoustic signal, is investigated.
Summary of current status: shi-jingpu-au00.pdf
Investigators: Resmi Kartha Komattil, Daniel E Hack
Advisor: Ashok Krishnamurthy
Sponsor: SBC and BETHA
Abstract: Elderly individuals with hearing loss have a great deal of difficulty conversing over the phone line. As people age, many develop sensorineural hearing loss that results in loss of sensitivity to high frequencies, especially at low intensity levels. Hearing aids help in overcoming this problem but many people find them prohibitively expensive or uncomfortable to use while conversing over the phone. In this project we are developing a speech preprocessor that can be used at the transmitting end of the phone line which will compensate for both phone line attenuation and an average hearing loss characteristic of hard-of-hearing listeners, in order to increase intelligibility. This project brings together individuals from the Franklin County Office on Aging, the Columbus Speech and Hearing Center, and the Departments of Speech and Hearing Science and Electrical Engineering at Ohio State.
Summary of current status: komattil-resmi-au00.pdf
Investigator: Benjamin J Buss
Advisor: Ashok K Krishnamurthy
Abstract: Testing is an expensive activity because it ties up valuable assets and delays further testing while waiting on results from earlier time consuming tests. This project aims to derive real-time or near real-time code from MATLAB/Simulink code.
Summary of current status: buss-benjamin-au03.pdf
Investigator: Jaeyoun Cho
Advisor: Ashok Krishnamurthy
Abstract: Speech enhancement is an important problem with applications in hearing aid design, speech recognition, speech coding etc. Spectral Subtraction and beamforming are exemplified as common methods for speech enhancement using single microphone and microphone array, respectively.
Summary of current status: Cho-Jaeyoun-au03.pdf
Authors: Jaeyoun Cho, Ashok Krishnamurthy
Title: Speech Enhancement with Beamforming Enhanced by Parametric Spectral
Subtraction
Conference: Acoustical Society of America First Pan-American/Iberian Meeting
on Acoustics, 2002
Status: Published
Authors: Jaeyoun Cho, Ashok Krishnamurthy
Title: Speech Enhancement using Microphone Array in Moving Vehicle
Environment
Conference: , 2003
Status: Published
Authors: Jaeyoun Cho, Ashok Krishnamurthy
Title: Speech Enhancement using Parametric Spectral Subtraction Combined
with Generalized Sidelobe Canceller
Conference: , 2003
Status: Published
Investigator: Ernesto O Machado-Colòn
Advisor: Ashok K Krishnamurthy
Abstract: Beat-tracking algorithms attempt to identify beats in music signals. The algorithm in development incorporates knowledge from psychoacoustics to detect periodicities in the signal's amplitude envelope, with transient and chord detection contributing to locating beat candidates. For comparison, data on human performance will be collected with a beat-labeling Matlab GUI.
Summary of current status: machado-ernesto-au03.pdf