Universities > United States Showing 1 - 20 of 1303 items
The University of Kansas (often referred to as KU or just Kansas) is a public research university with campuses located in Lawrence, Kansas City, and Overland Park, Kansas with the main campus being located atop Mount Oread in Lawrence. The University was founded in 1865 by the citizens of Lawrence under a charter from the Kansas Legislature. It claims the title of flagship university of the state of Kansas.
More information: http://en.wikipedia.org/wiki/University_of_Kansas
Python Object Serialization Pickling and Shelving
Vasudevan Muthukrishnan
Python Object Serialization Pickling and Shelving Vasudevan Muthukrishnan |
|
Beyond Genome Annotation -Characterizing Chromosome Features Terry Clark Assistant Professor Electrical Engineering and Computer Science The University of Kansas
Beyond Genome Annotation -Characterizing Chromosome Features Terry Clark Assistant Professor Electrical Engineering and Computer Science The University of Kansas2005 ITTC Research Review April 7, 2005 |
|
Productivity Tools: Make Vasudevan Muthukrishnan
Productivity Tools: Make Vasudevan Muthukrishnan |
|
e-LEARNING DESIGN LABORATORY Ed Meyen & John Gauch
Co-Directors of the e-Learning Design Lab
e-LEARNING DESIGN LABORATORY Ed Meyen & John Gauch Co-Directors of the e-Learning Design LabMission Plan Organization Discussion Demos |
|
An Introduction to the Information & Telecommunication Technology Center ITTC Technology Review May 2003
An Introduction to the Information & Telecommunication Technology Center ITTC Technology Review May 2003Victor S. Frost Director, Information & Telecommunication Technology Center Dan F. Servey Distinguished Professor of Electrical Engineering & Computer Science frost@eecs.ku.edu, 785-864-4833 |
|
Non-Contiguous Orthogonal Frequency Division Multiplexing Rakesh Rajbanshi and Gary J. Minden (rajbansh, gminden@ittc.ku.edu)
Non-Contiguous Orthogonal Frequency Division Multiplexing Rakesh Rajbanshi and Gary J. Minden (rajbansh, gminden@ittc.ku.edu)√ PAPR √ Error Robustness √ Overhead √ √ Throughput √ √ Spectrum Agility √ Synchronization NC-OFDM OFDM MC-CDMA Characteristics Introduction Qualitative Comparison * ‘√’ denotes the transceiver with best performance Fig. 2 NC-OFDM Transmitter Fig. 3 NC-OFDM Receiver Fig. 1 Schematic of spectrum usage BER Performance Evaluation Fig. 4 BER performance of NC-OFDM (solid lines) and MC-CDMA1 (dashed lines) transceiver Efficient Modulation Scheme Fig. 6 Mean execution times for 1024-point FFT employing the 3 FFT algorithms Fig. 5 FFT butterfly structure. A value of ‘0’ denotes a zero-valued subcarrier and ‘x’ denotes a data bearing subcarrier (a) AWGN Channel (b) Rayleigh Channel 1 Simulation for MC-CDMA Error Robustness is done by Chen Qi NC-OFDM is a variant of OFDM capable of Spectrum Pooling Can utilize RF environment measurements to avoid interference with existing systems and adjust to government regulations Can utilize non-contiguous bands collectively for high data rate communications |
|
SWAN: Sensor Webs as Agents who Negotiate Costas Tsatsoulis (PI) [tsatsoul@ittc.ku.edu], Ed Komp, Najla Ahmad, Chris Redford
SWAN: Sensor Webs as Agents who Negotiate Costas Tsatsoulis (PI) [tsatsoul@ittc.ku.edu], Ed Komp, Najla Ahmad, Chris RedfordVisualizer Agent Communication Protocol Example Scenario Visualization A visualization of a simulated area of land. Various layers display different types of information about the area. Ground Saturation Layer Colors represent the saturation percentages of the area. The pink square represents the current location of a roving sensor. Elevation Layer Colors represent varying elevations of the area. The areas which are shaded yellow are currently under observation by a satellite. Rainfall Layer Colors represent the current rate of rainfall. The red square represents the location of a sensor which measures this. Overview Agent 0 Measures: Rainfall Mobility: Immobile Agent 1 Measures: Ground Saturation Mobility: Ground Agent 2 Measures: Elevation Mobility: Orbit (Satellite) Registration When sensor agents enter the system, they register with the matchmaker. This gives the matchmaker their unique ID and allows it to assign and tell them their queue number. The agent then advertises its capabilities to the matchmaker so that it can match them with agents who need their capabilities. Capability Request When an agent requires a special sensing capability that it doesn’t have itself, it asks the matchmaker to recommend all of the agents it thinks can meet its capabilities. The agent then asks if one of these other sensing agents can perform the capabilities it needs. Negotiation Sometimes an agent will give a simple yes or no answer to t... |
|
Serially Concatenated Codes for Aeronautical Telemetry Kanagaraj Damodaran, Erik Perrins [esp@eecs.ku.edu]
Serially Concatenated Codes for Aeronautical Telemetry Kanagaraj Damodaran, Erik Perrins [esp@eecs.ku.edu]BER Performance of SCCC-SOQPSK-TG Motivation Proposed Solution BER Performance of SCCC-PCM/FM Bandwidth efficient serially concatenated coded (SCC) CPM techniques are developed for aeronautical telemetry. CPM waveforms like shaped-offset quadrature phase shift keying (SOQPSK) & pulse code modulation/frequency modulation (PCM/FM) can be viewed as inner codes due to their recursive nature. Efficient channel codes like convolutional codes & turbo-product codes can be used as outer codes. A family of high rate SCCs can be developed from basic SCC-CPMs to improve spectral efficiency. A coding gain of 4.6 dB at a code rate 7/8 A coding gain of 4.6 dB at a code rate 7/8 Forward error correction (FEC) schemes for aeronautical telemetry have received only preliminary attention to date. In recent years bandwidth efficiency has become a major concern in aeronautical telemetry. Moreover spectrum reallocations in 1997 prompted a search for more bandwidth efficient waveforms. Continuous phase modulation (CPM) is a constant envelope, power and bandwidth efficient digital modulation technique. |
|
Genetic Algorithms for Cognitive Radios Tim Newman (newman@ittc.ku.edu), Brett Barker (barkerb@ittc.ku.edu), and Gary Minden (gminden@ittc.ku.edu)
Genetic Algorithms for Cognitive Radios Tim Newman (newman@ittc.ku.edu), Brett Barker (barkerb@ittc.ku.edu), and Gary Minden (gminden@ittc.ku.edu)Fitness Function Genetic Algorithms Chromosome Parameters Fitness Function Solution ‘chromosomes’ are represented as binary strings composed of joined parameters Parameters Transmit Power Modulation Frame Size Channel Coding Rate Determine which chromosomes are “best” Goals are used to direct fitness scores Minimize BER Minimize Power Consumption Maximize Throughput Simple Fitness Function derived by inspecting the tradeoffs between the 2 parameters and the goals. Graphical Representation of fitness mapping for a given set of weights. The search space becomes large when using more parameters, which may span multiple network layers (PHY,MAC). Complex analysis will give us fitness functions that will allow a GA to take advantage of the parameters made available using a software defined radio architecture. The maximum point on this graph represents the optimal parameter settings for a given set of goals. Why do we even need a cognitive algorithm if we can see the maximum point? Genetic Algorithms (GA) provide a learning mechanism based on biological evolution. New decisions are made based on mutating and combining current decisions. At each step we have a population of decisions that we select to be most fit. Genetic Algorithms provide a parallel search over different parts of the problem space. |
|
KUAR WiMAX 802.16a OFDM Jordan Guffey
KUAR WiMAX 802.16a OFDM Jordan GuffeyOFDM Spectrum WiMAX OFDM KUAR Implementation The WiMAX standard 802.16a (2-11 GHz) supports an OFDM (Orthogonal Frequency Division Multiplexing) physical layer. OFDM allows much higher data rates over non-line-of-sight connections. One high-rate data stream is broken up into many lower-rate data streams. Each stream is assigned to a OFDM sub-carrier. This scheme mitigates frequency selective fading, a major bottleneck to wireless communications. Frequency domain view of a 4 carrier OFDM signal. Under ideal conditions, there is zero inter-carrier interference. Goal: Implement an adjustable OFDM transceiver in the KUAR (KU Agile Radio) that complies with the 802.16a OFDM PHY specifications. Challenges: Fitting all the required logic in the Virtex II FPGA. Complex error control coding scheme. OFDM extremely sensitive to timing and frequency offsets. Strategy: Implement the OFDM model in Simulink, then synthesize each block in VHDL. Each block will then be extensively tested in the FPGA with data generated from the Simulink model. Frequency selective fading with a single carrier system must be corrected by equalization. At very high data rates, equalization can become prohibitively complicated. Worldwide Interoperability for Microwave Access Designed to provide Metropolitan Area Access. Provides for wireless networking access at distances up to 50 km, rather than hundreds of meters with WiFi. MAN ( Metropolitan Area Network) rather than LAN ( Local Area Network ).... |
|
FAST: A ROC-based Feature Selection Metric for Small Samples and Imbalanced Data Classification Problems Xue-wen Chen (xwchen@ittc.ku.edu) and Mike Wasikowski
FAST: A ROC-based Feature Selection Metric for Small Samples and Imbalanced Data Classification Problems Xue-wen Chen (xwchen@ittc.ku.edu) and Mike WasikowskiExperiments Motivation Method: FAST Results Evaluate goodness of single feature linear classifiers along multiple possible thresholds Ranks individual features by area under ROC, comparison of TPR and FPR on each threshold Best features scores close to 1, worst features score close to 0.5 Compare FAST-selected features with RELIEF and correlation coefficient-selected features Train a linear SVM using selected features Use balanced 10-fold cross-validation to test effectiveness of selected features Repeat cross-validation trials 20 times Evaluate predictions by macro-averaged area under ROC across two microarray and two mass spectrometry data sets FAST-selected features beat the baseline with 30 features Correlation coefficient-selected features beat the baseline with 80 features RELIEF-selected features could not beat the baseline with <= 100 features Microarray and mass spectrometry data sets pose problems for traditional machine learning methods Small number of samples in data set (less than 100) Significantly less samples available in one class (have a disease) than the other (no signs of disease) Large number of features (6000 or more) with many being irrelevant to prediction task New approach: use feature selection metrics to identify those features with the most relevance to the problem Goal of using feature selection: improve classification performance on test data |
|
The SAFFIRE Algorithm for Spatio-Temporal Reconstruction of MEG/EEG Signals Dr. Shannon Blunt, Tszping Chan sdblunt@ittc.ku.edu
The SAFFIRE Algorithm for Spatio-Temporal Reconstruction of MEG/EEG Signals Dr. Shannon Blunt, Tszping Chan sdblunt@ittc.ku.eduImpact Overview Source AFFine Image REconstruction Results - ASSR Example Minimum mean square error approach Incorporate linear underlying forward model Leadfields matrix B Transform dipole activity into MEG sensor signal data Underdetermined model and extremely ill-conditioned Filter formulation: Iteratively update structured covariance matrix Rx Implement filter in the affine-transformed domain Matched filter initialization Ensure convergence to true solution Process multiple snapshots for SNR gain Advantages No prior statistical knowledge required Unprecedented spatio-temporal resolution Temporal signal correlation robustness Powerful bio-signal processing tool Successfully separated primary and secondary ASSR Previously unsolved problem in brain imaging community Patent-pending Auditory Steady-State Response Primary Dipole Index 1088 (4.7 cm) 7952 (4.4 cm) Secondary Dipole Index 613 (5.5 cm) 8570 (5.5 cm) Peak Delay 50 [ms] Dipole strength 30*10-9 [A m] Additive Noise Magnitude 10*10-15 [T] SNR 17 ~ 22 [dB] Red: Primary auditory dipole pair Green: Secondary auditory dipole pair Blue: Estimated dipole location Comparison - Mirrored Dipole Example Magneto- and Electro- encephalography (MEG/EEG) Neural-signals induced electromagnetic fields which are detectable outside of the brain Functional Neuroimaging Processing and analysis of sensor data Analyzes the functional purposes served by different regi... |
|
Comparative Analysis of EPC Compliant Devices Daniel D. Deavours (PI), Afzal Syed, Padmaja Yatham, Shilpa Sirikonda deavours@ittc.ku.edu
Comparative Analysis of EPC Compliant Devices Daniel D. Deavours (PI), Afzal Syed, Padmaja Yatham, Shilpa Sirikonda deavours@ittc.ku.eduTimelines RFID – The technology Alliance Lab Capabilities Tags, Readers and some results Frequency Dependence and radiation patterns Read rates in population Variance Performance near water and metal. Performance of devices in noisy environments Under Development Impact Abrasion Scraping Electrostatic Discharge (ESD) April 2004 RFID Alliance Lab established at the University of Kansas. Dec 2004 First Report titled “Performance Analysis of Commercially Available UHF RFID Tags” published May 2005 Second Report titled “UHF EPC Tag Performance Evaluation” published. July 2005 Started extensive research and development along with testing. Summer of 2006 Third report focusing on reader performance expected. Frequency Dependence and Radiation Patterns Tags Near Metal Tags in Population Variance among a single tag type A few Tags Thingmagic Reader Radio Frequency Identification (RFID) concept originated from World War II Automatic Identification (Auto-ID) of objects Non-line of sight identification Tracking products and Near-perfect supply chain Internet of things EPCglobal leads the development of industry-driven standards for RFID |
|
Concatenated Codes for Aeronautical Telemetry Kanagaraj Damodaran, Erik Perrins [esp@eecs.ku.edu]
Concatenated Codes for Aeronautical Telemetry Kanagaraj Damodaran, Erik Perrins [esp@eecs.ku.edu]Iterative Decoding Motivation Proposed Solution BER Performance High rate concatenated codes improves spectral efficiency. Concatenated Codes Serial concatenation of codes proved to be better in performance than parallel concatenation. In the above technique we concatenate an outer non-recursive encoder separated from an inner (CPM)modulator (which is viewed as a code) by an interleaver. Continuous Phase Modulation (CPM) is a constant envelope, power and bandwidth efficient digital modulation technique. As a common folk theorem states “All codes are good, except those that we know how to decode”. We use a novel, low-complexity, sub-optimum iterative decoding algorithm. The crux of the iterative decoding algorithm is a soft-input soft-output (SISO) a posteriori probability (APP) section. rate ½ concatenated codes with CPM provides approximately a gain of 2.5 dB over conventional codes. In the modern world of communication systems, channel coding has become an indispensable tool to satisfy power and bandwidth constraints. As made clear by Shannon, large coding gains can be obtained by encoding large blocks of information sequences. However spectrum being a limited resource, encoding and spectral efficiency prove to be conflicting requirements which represent a basic communications problem. |
|