Personal tools

2012 ARR Abstracts

Technical Program

  1. Biological and Chemical Detection Technologies
  2. New Trends in Information Processing
  3. Robust CAD
  4. Novel Lasers
  5. Advanced Biomedical Imaging
  6. New Coding Methods
  7. Cutting Edge Circuit Designs
  8. Innovative Devices
  9. Novel Communication Algorithms
  10. New Methodologies for Circuits
  11. High Performance Receivers and Antennas
  12. Advances in Devices and Materials
  13. Advanced Inference Algorithms
  14. Next-Generation DSP Architectures

Session 1: Biological and Chemical Detection Technologies

"ZIF-coupled Microresonator for Highly Sensitive and Selective Gas Detection"
Yongha Hwang and Rob Candler
Abstract: We introduce a zeolitic imidazolate framework (ZIF)-coupled resonant gas sensor whose sensitivity shows an improvement up to 78 times over silicon resonant sensors with identical dimensions while additionally utilizing the inherently selective adsorption properties of ZIFs. In order to achieve high sensitivities, a resonator needs to have a high surface-to-volume ratio. In this study, we demonstrate a microresonator coupled with ZIF crystals, which provide a level of surface area that was previously unachievable, as well as the potential for inherent selectivity.
"Wide-field Lensless On-Chip Fluorescence Microscopy for High-throughput Screening of Bio-Chips" 
Ahmet F. Coskun, Ting-Wei Su, Ikbal Sencan, David Herman, YeongSeok Suh and Aydogan Ozcan
Abstract: On-chip fluorescence microscopy is an emerging platform, achieving < 4 µm spatial resolution with multi-color and multi-layer imaging capabilities, that enables high-throughput monitoring of bio-chips over a wide field-of-view of e.g., >9-18 cm2 without the use of any lenses, thin-film filters or mechanical scanners. In this talk, we will review the recent advances in wide-field lensless fluorescence microscopy and also discuss some of its unique abilities toward high-throughput screening applications, including rare-cell imaging, on-chip cytometry and as well as small animal model research.
"Cell-phone based Rapid-Diagnostic-Test Reader Platform for Real-time Mapping of Infectious Diseases" 
Onur Mudanyali, Stoyan Dimitrov, Uzair Sikora, Swati Padmanabhan, Isa Navruz, and Aydogan Ozcan  
Abstract: Rapid-diagnostic-tests (e.g., immunochromatographic-tests) take center stage in the surveillance of epidemics especially in the remote parts of the world, offering significant advantages over conventional approaches (e.g., clinical examination). In order to enable spatio-temporal mapping of infectious diseases and provide accurate digital reading of rapid-diagnostic-tests, we recently developed a universal RDT reader attachment installed on a cell-phone that can automatically digitize and evaluate various diagnostics test results. A custom-developed cell-phone application generates semi-quantitative test evaluation reports and also connects the RDT reader to a central database/server to store and display spatio-temporal statistics for the prevalence of various infectious diseases in real-time.

Session 2. New Trends in Information Processing

"Balancing Behavioral Privacy and Information Utility in Sensor Data Flows" 
Supriyo Chakraborty, Kasturi Rangan Raghavan and Mani B. Srivastava
Abstract: Smart phones have enabled the growth of context-aware applications providing services tailored to user contexts. These contexts typically define the personal, social, work and urban spaces of an individual and are derived from the underlying sensor measurements. The shared context streams 
often embed in them personal information raising serious privacy concerns.  In this talk, I will describe the design and implementation of a context-aware privacy preserving mobile system on the Android OS. Specifically, I will focus on the platform level changes, and the privacy-rule specification module of the framework. I will conclude with possible extensions to the current framework.
"On-line Activity Classification Based on Contexts with Portable Device for Wireless Health" 
Chieh Chien, Hua-I Chang and Greg Pottie
Abstract: Obtaining accurate measurements of human activities is important for a broad set of health applications. We propose a context-based hybrid decision tree classifier with a real-time portable solution for reliably classifying daily life activities and providing instant feedback. At first, to determine user contexts, we utilize tablet sensors to collect Wi-Fi fingerprint and capture movement. Then, we select different types of hybrid decision tree classifiers based on estimated human context. The tree classifier can flexibly implement different decision rules at its internal nodes, and can be adapted from a population-based model when supplemented by training data for individuals. In addition, with the introduction of portable devices, the users can receive instant feedback of their current mobility status.
"Optimal Relay Deployment for Robust Multicast Scheduling in Cellular Wireless Networks" 
Hung-Bin Chang and Izhak Rubin
Abstract: We develop adaptive-rate relay-aided scheduling algorithms for multicasting critical messages in wireless cellular networks. We consider multi-phase coloring oriented scheduling algorithms, involving dynamic coordination among base station and relay nodes.  We compare the system's spectral efficiency under various reuse levels.  We determine the optimal locations at which relay nodes should be placed in each cell.  We carry out performance analyses to establish, for each scheduling scheme and multi-hop routing configuration, the system's performance and the contributions made by the use of relay nodes.  We also study the best mechanisms to be used under the failure of base station nodes.   

Session 3. Robust CAD

"Two Efficient Ways to Monitor the Maximum Operating Frequency of a Chip"
Liangzhen Lai and Puneet Gupta
Abstract: In this presentation, we are going to talk about two proposed methodologies to identify chip maximum operating frequency. The first method synthesizes a few replica ring-oscillators based on the technology and design properties. Experiments show that our delay estimation method using multiple ring-oscillators reduces chip delay overestimation by up to 25% compared to use of a single one.
The second method is to monitor in-situ timing slacks by our novel monitor insertion flow. Preliminary results show that our method can reduce the total number of monitors required of by more than an order of magnitude.
"Modeling and Optimization for Circuit Resilience" 
Fang Gong and Lei He
Abstract: As the semiconductor industry scales down to the nanometer regime, it has become increasingly challenging to predict circuit behavior and failure probability in the presence of process variations. In our previous work, we first proposed an efficient yield analysis algorithm based on global search to estimate the yield rate under process variation accurately; In addition, a stochastic algorithm based on point estimation method has been developed to estimate the arbitrary probability distribution (e.g., PDF and CDF) for circuit performance merits. Moreover, we proposed an efficient algorithm to estimate extremely small failure probability. Extensive experiments have validated our proposed approaches. 
"Soft Error Mitigation for FPGA Based Systems" 
Ju-Yueh Lee and Lei He
Abstract: Field programmable gate arrays (FPGAs) use primarily SRAM cells to implement field programmability. However, FPGA configuration SRAM cells are vulnerable to Single-Event-Upset (SEUs). These upsets are soft errors, but have permanent impact until the FPGA is re-programmed. We have developed a number of techniques compatible with existing FPGA chips and synthesis flow, including in-place logic synthesis techniques. We are able to improve MTTF by 2X with little or no overhead for circuit area, performance and power, and with no change on design closure. We have also developed a soft error modeling and validation platform using computer-based fault injection. 

Session 4. Novel Lasers

"Developing Novel Computational Methods for Designing Compact Plasma-Based Accelerators for use in Next Generation Light Sources and Linear Colliders" 
Peicheng Yu and Warren Mori
Abstract: Plasma-based acceleration has the potential for being a technology for building compact accelerators for diverse applications ranging from medicine, material science, and high energy physics. We describe a  novel computer code that models laser wakefield acceleration in the Lorentz boosted frame. We show that our method overcomes some numerical instabilities that have been a challenge in the past.  We show how this approach reduces the computer resources needed to model laser wakefield accelerator stages and discuss areas for future work
"Generation of Ultra-fast CO2 Laser Pulses using Self-Phase Modulation in a Noble-Gas-Filled Hollow Waveguide" 
Jeremy Pigeon, Sergei Tochitsky, Chao Gong and Chan Joshi 
Abstract: There has been a renewal of interest in the development of high power, broadband, mid-IR laser sources driven by applications in remote sensing, X-ray generation and advanced acceleration, for which a CO2 laser is an obvious candidate. Here we discuss the development of a 1 Hz repetition rate, 20 GW, 3 ps CO2 laser which relies on ac Stark Broadening for final amplification. We then propose an experiment to guide these pulses in a noble-gas-filled hollow glass waveguide. Simulations of this interaction indicate that such a scheme may be used to increase our peak power by a factor of 5-10. 
"Stabilizing the Periodic Oscillations of Semiconductor Lasers for Precision Photonic Microwave Generation" 
Mohammad AlMulla and Jia-Ming Liu
Abstract: Photonic microwave or millimeter wave (MMW) sources that produce highly stable and broadly tunable microwave frequencies of low phase noise are anticipated for many applications ranging from broadband wireless access networks and satellite communication systems to emerging broadband photonics-based phased-array antennas and radars.
Many methods are used to stabilize the microwave generated from optically injected semiconductor lasers such as optoelectronic feedback, double locking, or by a phase-locked loop. Currently, the narrowest microwave linewidth achieved is on the order of millihertz.
In this talk, we investigate the attributes of specific operational points within the period-one oscillation region of the nonlinear dynamics induced by an optically injected semiconductor laser. These points have low noise and are insensitive to jitter induced from fluctuations of the operating conditions. Therefore, these operating points can further stabilize the generated microwave frequencies for microwave photonic applications.

Session 5. Advanced Biomedical Imaging

"Lensfree Holographic Microscopy for Automated On-Chip Semen Analysis" 
Ting-Wei Su, Anthony Erlinger, Derek Tseng, and Aydogan Ozcan
Abstract: We demonstrate automated semen analysis using lensfree on-chip holography. This computational microscopy scheme does not require any lenses, lasers or other bulky optical components to achieve phase and amplitude imaging for thousands of sperms over ~24 mm^2 field-of-view with a numerical aperture of ~0.2. Such a simple and automated semen analysis platform that can investigate both sperm concentration and motility with a high throughput is especially valuable for male fertility testing, livestock breeding, as well as for statistical analysis of sperm dynamics.
"A Wireless Handheld Biomedical Imaging System for Monitoring Pressure Ulcer Development" 
Frank Wang, Yeung Lam, Alireza Mehrnia, Barbara Bates-Jensen, Majid Sarrafzadeh, and William Kaiser  
Abstract: Millions of patients each year suffer from pressure ulcers leading to severe complications and over 60,000 mortalities. Early detection of pressure ulcer occurrence can reduce risk. However, the only available detection method, based on visual inspection of tissue by an evaluator, is inconvenient for the patient and requires an expert evaluator.  This paper describes the development of a compact wireless handheld instrument that introduces the first evidence-based method for detection of pressure ulcers.  This applies methods to characterize the dielectric properties of tissue and detect subepidermal moisture that appears in the presence of pressure ulcers. The efficacy of this device has been directly demonstrated by trials based on direct measurement of induced subepidermal moisture. Its efficacy for supporting patients in a healthcare facility has been based on a second trial that includes over 700 measurements of patient anatomical sites. The fundamental principles, design requirements, implementation and experimental evaluation are reported here.
"Quasi-Fluorescence Lifetime Detection of Brain Tumor and Tissue Structure for Intraoperative Use" 
Asael Papour, Zach Taylor, Maie St. John, William Yong, Oscar Stafsudd, and Warren Grundfest
Abstract: This research employs technological innovations by using light emitting diodes in conjunction with novel fluorescence lifetime contrast generation algorithms for the detection and delineation of different tissue types and structures. The goal of this study is to evaluate the efficacy of this technology for tissue characterization. The fast acquisition rate and agile performance can be applied into an intraoperative imaging device for the detection of brain and head and neck tumors. Ongoing research examining large numbers of samples will determine the specificity and sensitivity of this method and its role for intraoperative decision-making.
"Perfusion Oxygenation Monitor: Wearable Continuously Vigilant Perfusion Imaging" 
Bijian Mapar and William Kaiser
Abstract: Rapidly assessing blood perfusion in tissue is an urgent need in healthcare, enabling diagnosis of circulatory disorders, wound conditions, and characterization of treatment outcome.  These high risk conditions require continuous monitoring to detect and treat appropriately, which is currently not offered by existing solutions.  In order to fill this crucial gap in healthcare a wireless health system, the Perfusion Oxygenation Monitor (POM), is under development.  This system utilizes sensor diversity by providing a multispectral optical array with distributed sensors scheduled to provide continuous monitoring of high risk tissue.  Inducing perfusion change through thermoregulatory vasodilation, perfusion assessment has been experimentally verified.

Session 6. New Coding Methods

"Dynamic Threshold Schemes for Multi-Level Nonvolatile Memories" 
Frederic Sala, Ryan Gabrys, and Lara Dolecek
Abstract: Nonvolatile memories, including flash memories, are an increasingly popular storage medium. In nonvolatile memories, reading stored data is done through the use of predetermined fixed thresholds. However, due to problems affecting such memories, including voltage drift and overwriting, fixed threshold usage often results in significant asymmetric errors.The notion of dynamic thresholds was introduced to combat these problems. In this talk, we explore the use of dynamic thresholds for multi-level cell (MLC) memories. We describe a high-performance general scheme to compute and apply dynamic thresholds. Finally, using this scheme, we develop asymmetric error-correcting codes well suited to nonvolatile memories.
"Performance Bounds for Analysis of Rate-Compatible Sphere-Packing with Numerous Transmissions" 
Tsung-Yi Chen, Adam R. Williamson, Dariush Divsalar, and Richard D. Wesel
Abstract: Recent results by Chen et al. and Polyanskiy et al. explore using feedback to approach capacity with short blocklengths. This paper explores Chernoff bounding techniques to extend the rate-compatible sphere-packing (RCSP) analysis proposed by Chen et al. to scenarios involving numerous retransmissions and different step sizes in each incremental retransmission.  Williamson et al. employ exact RCSP computations for up to six transmissions.  However, exact RCSP computation with more than six retransmissions becomes unwieldy because of joint error probabilities involving numerous chi-squared distributions. This paper explores Chernoff approaches for upper and lower bounds on the error probability to provide support for computations involving more than six transmissions.. 
"High Throughput with Low Latency: The Analysis and Design of Rate-Compatible Codes for Gaussian Channels with Feedback" 
Adam R. Williamson, Tsung-Yi Chen, and Richard D. Wesel
Abstract: Feedback can dramatically improve the throughput of communication systems with low latency (i.e., average blocklength), though it does not increase the asymptotic (Shannon) capacity of memoryless channels. We present a feedback-based incremental redundancy scheme which uses a rate-compatible sphere-packing analysis and applies recent results in finite blocklength information theory to demonstrate the throughput and latency benefits of feedback in Gaussian channels. A two-phase communication and confirmation scheme is implemented in order to reduce the probability of undetected error. Simulation results of practical convolutional codes are provided.
"Coding for Multilevel Write-Once Memories" 
Ryan Gabrys and Lara Dolecek
Abstract: A Write-Once Memory (WOM)-code is a coding scheme that allows information to be written in a memory block multiple times, but in a way that the stored values are not decreased across writes. This work studies non-binary WOM- codes with potential applications to Flash memory. We begin by presenting some simple constructions for non-binary WOM- codes. Then, some of the necessary and sufficient conditions of capacity-achieving WOM-codes are studied. Aided by the resulting insight into the code structure, a new class of two- write non-binary codes, called Spider Codes, is introduced. In addition to unrestricted rate per generation, constructions and bounds of fixed-rate non-binary codes are also considered. In many instances, the proposed code constructions provide the highest known sum-rates for the non-binary WOM.

Session 7. Cutting Edge Circuit Designs

"A Multichannel, Multicore mm-Wave Clustered VCO with Phase Noise, Tuning Range, and Lifetime Reliability Enhancements" 
Farid Shirinfar, Med Nariman, Tirdad Sowlati, Maryam Rofougaran, Reza Rofougaran, and Sudhakar Pamarti
Abstract: A VCO design technique with phase noise, tuning range, and reliability enhancements is proposed.  A proof-of-concept design targeting the WiGig protocol is presented.  Each cluster of VCOs covers one channel resulting in better phase noise performance.  Multicores of VCOs are combined to further enhance phase noise and combat the voltage swing reliability issues. The VCO achieves a phase noise of-101.8dBc/Hz at 1MHz offset with over 12.6% tuning range (50.7GHz 57.5GHz) and an FOM of -182dB/Hz. 
"A Fully Integrated 22.6dBm 60GHz Power Amplifier in 40nm CMOS" 
Farid Shirinfar, Med Nariman, Tirdad Sowlati, Maryam Rofougaran, Reza Rofougaran, and Sudhakar Pamarti
Abstract: A fully integrated 60GHz CMOS PA with a PSAT of 22.6dBm is presented.  PSAT remains above 17dBm even with a VDD of 0.7V.  The PA consists of a pre-PA followed by eight 4-stage transformer-coupled PA chains with a total gain of 29dB.  The final combining network combines eight 50ohm differential inputs into a 50ohm single-ended output with 3dB loss.  Voltage related reliability issues are addressed by novel dual input/dual output transformers. 
"A 10-Bit 1-GHz CMOS ADC with FOM=70 fJ/Conv." 
Sedigheh Hashemi and Behzad Razavi
Abstract: A pipelined ADC incorporates a precharged resistor-ladder DAC in amulti-bit front-end, achieving fast settling and allowing calibrationof both dynamic and static gain errors. Using simple differential pairs with a gain of 5 as op amps and realized in 65-nm CMOS technology, the 10-bit ADC consumes 36 mWat a sampling rate of 1 GHz and exhibits an SNDR of 52.7 dB at an input frequency of 490MHz.
"A 10-Bit 4-GHz CMOS ADC"
Hegong Wei and Behzad Razavi

Abstract: A 4-channel time-interleaved (TI) 10-bit 4GS/s pipelined Analog-to-Digital Converter (ADC) is designed and fabricated on 65nm CMOS. To improve the dynamic performance at high input frequencies, the background timing skew calibration is employed. The timing skew is detected in digital domain through the proposed calibration algorithms and minimized by adjusting digitally controlled delay lines. The proposed calibration algorithm and circuitry provides a powerful correction, allowing input rate over Nyquist and enhancing SFDR by up to 40dB. The power consumption is 180mW from 1.2V/1.4V supplies and the ADC core occupies an area of 1.2mm2.

Session 8. Innovative Devices

"InAs/ AlAsSb Self-Assembled Quantum Dots for Intermediate Band Solar Cells" 
Meng Sun, Paul J. Simmonds, Ramesh Babu Laghumavarapu, Andrew Lin, Baolai Liang, and Diana Huffaker
Abstract: In order to realize solar cells with higher efficiencies, interest has recently focused on intermediate band solar cells (IBSCs) based on the use of quantum dots (QDs). IBSCs benefit from high photocurrents generated due to the absorption of sub bandgap photons between the valence band (VB) and intermediate band (IB) and the intermediate band and conduction band (CB). Our work centers on InAs QDs surrounded by AlAsSb barriers lattice matched to InP. The InAs/AlAsSb system has several advantages over more traditional QD systems such as InAs/GaAs, InGaAs/GaAs, and GaSb/GaAs. By engineering dot and barrier materials, InAs/AlAsSb system has the potential to a) achieve a negligible valence band offset and b) obtain appropriate bandgap alignments to form intermediate bands. 
"Device Modeling for Tunneling Transistors"
Andrew Pan and Chi On Chui
Abstract: Tunneling transistors (TFETs) are being intensely studied because of their potential for low voltage and power operation. Analytical models for emerging devices are crucial for physical understanding as well as circuit evaluation; however the few TFET models which exist in the literature are limited in scope and accuracy. In this talk I will present our work on TFET modeling, accounting for major physical effects, differing geometries, and design parameters. We demonstrate model agreement with a wide range of TCAD simulations as well as experimental data. Design insights from the model as well as areas for continued work are discussed.
"Reducing Electrostatic Screening in Field-Effect-Transistor-Based Biosensors" 
Kaveh Shoorideh and Chi On Chui
Abstract: In FET-based biosensors, the gate stack is replaced by an electrolytic solution containing biological molecules of interest. The charges of these molecules are used to gate the underlying FET. Ions in the electrolyte screen the electric field emanating from the biomolecular charges, reducing
the sensitivity of the biosensor. In most cases, researchers resort to using buffer solutions with ionic strengths far below those of physiologically relevant environments in order to circumvent the screening problem. In this talk, we will show how screening can be reduced simply by choosing the right sensor geometry.
"Tuning the InGaAs Alloy Composition in Nanopillars with Selective-Area Pattern Geometry" 
Joshua Shapiro, Adam Scofield and Diana Huffaker
Abstract: Catalyst-free selective-area-epitaxy (SAE) is a powerful technique for growing uniform arrays of nanowires, and accurately controlling their location, diameter, and height.   The precise positioning, and excellent uniformity of SAE nanopillars makes this growth technique highly advantageous for device fabrication.   In this work, GaAs nanopillars with ~100nm long axial InGaAs inserts are grown by MOCVD via catalyst-free selective-area-epitaxy.  We show how the alloy composition of the InGaAs region depends critically on the pitch and diameter of the selective-area pattern geometry.  The PL emission varies based on pattern geometry from 1.05 µm to 1.25 µm corresponding to indium fractions from 0.15 to 0.3.  This indium enrichment is explained by a relative change in the effective flux of indium and gallium controlled by the selective-area pattern.  This work has potential applications in 4th generation multi-spectral imagers that seek to engineer spectral information directly into sub-wavelength pixels on a focal plane array.

Session 9. Advanced Algorithms for Speech and Signal Processing 

"Cramer-Rao Bounds for Joint RSS/DoA-Based Primary-User Localization in Radio Networks" 
Jun Wang, Jianshu Chen, and Danijela Cabric  
Abstract: Knowledge about the location of licensed primary-users (PU) could enable several key features in cognitive radio (CR) networks including improved spatio-temporal sensing, intelligent location-aware routing, as well as aiding spectrum policy enforcement. In this paper we consider the achievable accuracy of PU localization algorithms that jointly utilize received-signal-strength (RSS) and direction-of-arrival (DoA) measurements by evaluating the Cramer-Rao Bound (CRB). Previous works evaluate the CRB for RSS-only and DoA-only localization algorithms separately and assume DoA estimation error variance is a fixed constant or rather independent of RSS. We derive the CRB for joint RSS/DoA-based PU localization algorithms based on the mathematical model of DoA estimation error variance as a function of RSS, for a given CR placement. The bound is compared with practical localization algorithms and the impact of several key parameters, such as number of nodes, number of antennas and samples, channel shadowing variance and correlation distance, on the achievable accuracy are thoroughly analyzed and discussed. We also derive the closed-form asymptotic CRB for uniform random CR placement, and perform theoretical and numerical studies on the required number of CRs such that the asymptotic CRB tightly approximates the numerical integration of the CRB for a given placement.
"Relay-Aided Failover Using Adaptive Unicast" 
Choo Chin Tan and Izhak Rubin 
Abstract: We develop adaptive power and rate scheduling algorithms with relay-aided failover schemes for unicasting messages to mobile clients in wireless cellular networks. Upon the failure of base stations, the failover schemes reconfigure the unicast schedule and packet distribution routes to provide best coverage to mobile clients in the failed cells while assuring a high ensuing throughput rate performance. We study the effectiveness of using relay stations to support the unicast transmissions under pre-failure and post-failure scenarios. We show that our failover scheduling algorithms are effective in adapting to base station failures, limiting the performance degradation incurred.
"Optimizing Video Transmission in 4G LTE Networks" 
Zhuang Jin, Jianwen Chen and John Villasenor
Abstract: With the continued global rollout of LTE cellular networks, the infrastructure to support anytime, anywhere delivery of high quality wireless video is finally maturing. However, there still remain significant open questions regarding the impact of protocol choice on video quality. In particular, while both Real Time Streaming Protocol (RTSP) and HyperText Transfer Protocol (HTTP) Streaming offer a mechanism to support video delivery, they involve different tradeoffs and optimization opportunities.
We have developed a 4G LTE network modeling environment to investigate these issues, and present results and comparisons relating to upload and download performance at a variety of bit rates and network conditions.
"Soft-Decision Maximum-Likelihood Trellis-Based Decoding of Linear Block Codes for Short Blocklength Codes"
Kasra Vakilinia and Richard D. Wesel
Abstract: Based on recent work by Polyanskiy et al. and by Chen et al., there is new interest in approaching capacity with short block lengths using rate-compatible codes with short block lengths that perform close to the sphere-packing bound.  Some of the best short block length codes are linear block codes that are constructed algebraically and hence are most commonly decoded using algebraic decoding techniques.  However, soft decision decoding is required for such codes to approach capacity with low latency using feedback.  In this project, the performance of soft-decision maximum-likelihood trellis-based decoders for certain linear block codes on an AWGN channel is evaluated.  The results are compared with other commonly used algebraic decoding algorithms. Improvements in error performance and coding gain are obtained by using the Viterbi algorithm for BCH codes in comparison with syndrome and Berlekamp–Massey decoding algorithms. 

Session 10. New Methodologies for Circuits

"Design-Aware Mask Inspection" 
Abde Ali Kagalwalla, Puneet Gupta, Chris Progler and Steve McDonald
Abstract: Mask manufacturing is the biggest contributor to the exploding cost of leading-edge semiconductor manufacturing. In this work, we propose techniques to reduce the cost of mask inspection, which can take up as much as 40% of mask manufacturing cost, by increasing its design awareness. This is accomplished by first locating non-functional features in a circuit layout, and using that information along with the timing information of the design to assign criticality to different layout shapes. This information can be exploited by mask inspection tools to significantly reduce defect review time and first pass yield of masks (20%-point for a critical poly silicon mask).
"New Approach to Multi-Stage Directional Filter Based on Band-Reject Filter Design" 
Jim S. Sun, Humberto Lobato-Morales, Alonso Corona-Chavez, and Tatsuo Itoh
Abstract: A new structure for realizing multi-stage directional filter is proposed in this paper. Instead of relying on one-stage directional filter as the building block, this approach relies on stand-alone band-reject filters. In this paper we investigate the viability of this idea by developing a three-stage directional filter prototype with elliptic response and 5.5% bandwidth. It is demonstrated that the final directional filter response is very close to the constituting band-reject filter, hence allows us to synthesize the directional filter with various kinds of responses in a more precise way based on the rich legacy from the filter community.
"Analysis of Metastability in Pipelined ADCs" 
Sedigheh Hashemi and Behzad Razavi
Abstract: Comparator metastability in a pipelined ADC can lead to an occasional large ADC error codes causing bit error rate (BER) in digital communication systems or instrumentation applications. Conventional metastability analyses do not sufficiently predict the associated error in a pipelined ADC.  In this work, we present a comprehensive analysis of the comparator metastability effects in a pipelined ADC environment and propose a method to precisely predict the error for a given pdf of the input signal. 
"Hybrid Combination of Dual Band Isolation Circuits Based on Conventional and CRLH Transmission lines for Triplexers" 
Hanseung Lee and Tatsuo Itoh
Abstract: Combination of two dual band isolation circuits using CRLH TLs and one dual band isolation circuit based on conventional TLs can form a triplexer. Since it is not necessary to modify filters and there is no limitation to choose filters in the design concept, commercial filters can be used for a triplexer. Also, the design process is simple and straightforward. For verifying the design concept, a triplexer, which has 881.5MHz, 1575MHz, and 1842.5MHz center frequencies, is designed. The simulated and measured results show that the insertion loss is less than 2dB, the return loss is larger than 14dB, and the isolation is larger than 32dB. 

Session 11. High Performance Receivers Antennas 

"Advances in CMOS THz Imaging"
Adrian Tang and Frank Chang
Abstract: Sub-mm-wave and mm-wave based imaging has recently gained interest for security screening, contraband detection, gesture recognition, and bio-imaging. Underlying all these applications is the need for transmitting and detecting high carrier frequencies to support high detection resolutions at stand-off distances of several meters. In this talk we review CMOS receivers based on super-regeneration that were developed at UCLA and operate up to 495 GHz. We will also discuss a 150 GHz integrated imaging radar system which can capture 3D images of objects at several meters of stand-off distance. 
"A High Gain Active Antenna Array using Dual-Fed Distributed Amplifier-Based CRLH-Leaky Wave Antennas" 
Chung-Tse Michael Wu and Tatsuo Itoh
Abstract: In this work, we propose a novel high gain active antenna array using dual-fed distributed amplifier (DA)-based composite right/left-handed (CRLH) leaky wave antennas (LWAs). Operating around 2.4GHz ISM-band, the active antenna array can achieve 16dBi gain at maximum, which has
around 7dB gain increase compared with the single element dual-fed DA-based active antenna. The proposed active antenna array is also capable of beam-scanning, with respect to the frequency of the pass band of the antenna.
"A 12-mW CMOS Receiver for IEEE 802.11a Applications"
Ali Homayoun and Behzad Razavi  
Abstract: While there are low-power ADCs and frequency synthesizers, the RF front-end and the filters of the receivers still consume relatively large amount of power. For example, state of the art 802.11a receivers consume about 70 mW or more. In this work, there is no static power consumption in the front-end since LNA is removed and passive mixers are used. The baseband filters are designed to have low noise and high blocker handling capability. The receiver consumes only 12 mW including the dynamic power of the LO buffers that drive the mixers.
"Comparison of Nature-Inspired Optimization Techniques for Antenna Applications in Communications"
Joshua Kovitz and Yahya Rahmat-Samii
Abstract: Nature-inspired optimization techniques encompass a wide variety of contemporary optimization routines, and two established algorithms have surfaced due to their strengths and advantages over other algorithms. They include the Particle Swarm Optimization (PSO) technique and the Covariance Matrix Adaptation Evolution Strategies (CMAES) technique. This research aims to compare the algorithms for resource-limited optimization problems, due to the heavy computational burden of antenna simulations. Several antennas are optimized for possible use in newly developing systems such as cognitive radio. The optimization results are compared, and they demonstrate the robustness of the algorithms and the designs provided. 

Session 12. Advances in Devices and Materials

"Graphene Undulators as Terahertz Radiation Sources" 
Aryan Navabi and Kang L. Wang
Abstract: We propose a graphene-based device as a source for terahertz radiation by utilizing graphene’s high electron velocity.  Terahertz radiation sources have been long studied and developed due to their application in security screening, communications, biological and medicine sciences.  However, there has been a lack of radiation sources in the terahertz regime at room temperature.  A graphene-based THz source can be realized by creating periodic wrinkles in graphene to make a graphene undulator, where charge carriers act as radiation source through cyclotron radiation.  Since the radiation is not due to band-to-band transitions, the device can be operational at room temperature.
"Distributedly Modulated Capacitors for Integrated Circulators" 
Shihan Qin and Yuanxun Ethan Wang
Abstract: The requirement of using ferrite magnetic material for non-reciprocal microwave components like circulators prevents the design from being small, lossless and broadband especially at the lower end of the microwave frequency spectrum. Distributedly Modulated Capacitors (DMC), employing only transmission lines and time-varying capacitors, provides a novel realization of lossless and broadband non-reciprocity with additional advantages including compatibility with modern integrated circuit technology. A prototype made on Rogers circuit board has verified the theory's prediction of the performance, which is capable of achieving >10dB isolation between transmitting and receiving path and <5dB receiving insertion loss for frequencies of 400MHz to 1.8GHz. 
"Crystal Shape Modeling and Control in Protein Crystal Growth" 
Michael Nayhouse and Panagiotis D. Christofides
Abstract: Proteins play a key role as therapeutics in a number of diseases and protein crystallization is a central activity in the pharmaceutical industry. Protein crystals, usually produced through a batch crystallization process, are desired to be of high quality, of desired shape, and within a narrow size and shape distribution range. Motivated by the above considerations, the present work focuses on the modeling and control of protein crystal shape. The model protein used for this work is the tetragonal hen egg white lysozyme. The growth of an individual lysozyme crystal is modeled via kinetic Monte Carlo (kMC) simulations comprising adsorption, desorption, and migration events on the (110) and (101) faces, which are assumed to be independent. The expressions for the rate equations for each event type are similar to those of Durbin and Feher. Extensive testing of the system parameters indicates crossover behavior between the growth rates of the two faces [i.e., (110) and (101)], a fact that has also been observed experimentally. A nonlinear algebraic equation that relates the steady-state growth rate ratios between the (110) and (101) faces, the temperature and concentration, is derived from the kMC simulation data. This nonlinear equation is then utilized by a model predictive controller which regulates the protein crystal to desired shapes subject to manipulated input constraints. The proposed method is shown to successfully regulate protein crystal shape, ranging from equidimensional to more elongated type of structures, in the presence of arbitrary variations of the protein concentration.
"Polarization Friendly Retrodirective Array" 
Brandon Jun Choi and Tatsuo Itoh
Abstract: A retrodirective array (RDA) capable of retransmitting predictable (orthogonal) polarization with respect to any received polarization state is presented. Generally, RDA has the unique ability to backscatter the received signal back to the interrogator without prior knowledge of the source location. In addition to this unique feature, the proposed system can receive any polarization and always retransmit the signal that is orthogonally polarized from its received polarization state back to the interrogator. The added feature helps to maintain a more secure communication link between the RDA and the interrogator by mitigating the polarization mismatch loss.

"Radiation and Beam Engineering of Terahertz Quantum-Cascade Lasers" 
Philip W. C. Hon, Amir A. Tavallaee, Zhijun Liu, Benjamin S. Williams, Tatsuo Itoh
Abstract: Terahertz quantum-cascade lasers based on the metal-metal waveguide configuration are promising sources of far-infrared radiation. However, the devices exhibit poor beam quality due to its subwavelength radiating apertures. To tailor the beam pattern of such devices, we have proposed the use composite right/left handed transmission lines metamaterial antennas. We have demonstrated an active leaky wave antenna and terahertz passive transmission-line metamaterials based upon metal-metal waveguide technology. We have also developed a model for radiation properties of these antennas based upon the cavity antenna model. We are able to predict the radiative loss, far-field polarization and far-field beam patterns of both conventional THz QC-laser metal-metal waveguide devices, and metamaterial leaky-wave antennas. The model's polarization predictions are verified experimentally by angle-resolved reflectivity spectroscopy of a composite right/left-handed transmission-line waveguide metasurface.

Session 13. Advanced Inference Algorithms

"Automated Inference of Political Group Trajectories in Social Media" 
Roja Bandari, Hazhir Rahmandad, and Vwani P. Roychowdhury
Abstract: We devise an automated and un-supervised methodology to summarize participation dynamics in social news websites. We apply this methodology to data that spans more than four years and that deals with pressing and potentially divisive political issues. Our approach avoids the challenge of meticulous text-based content processing by determining evolving groups of users with similar interest patterns. We show that political dynamics in these informal online institutions can be effectively and automatically indexed and inferred at multiple scales, and that in spite of their magnitude and largely unregulated nature, many of these sites exhibit coherent, albeit complex, dynamics.
"Automatic Estimation of Speaker Height From Speech Signals"
Harish Arsikere and Abeer Alwan
Abstract: Speech is produced when airflow from the lungs is driven by the subglottal system, modulated by the vocal folds, and filtered by the vocal tract. The resonances of the subglottal system (SGRs) have only been recently studied by speech scientists. Our research focuses on
studying the role of SGRs in speech production, estimating SGRs from speech signals, and applying them to problems in speech processing. Our recent studies revealed that SGRs are related to speaker height and this information led to the development of an automatic algorithm for estimating speaker height (to within 5.3 cm, on average) from speech signals.
"Convex Optimization Methods for Topology Selection in Graphical Models" 
Jinchao Li and Lieven Vandenberghe
Abstract: A Gaussian graphical model is a graph representation of the conditional independence relations between components of a Gaussian random vector. A fundamental problem in the construction of large graphical models from observed data is the selection of the topology of the network, a problem often referred to as the covariance selection problem. Recently developed methods for sparse covariance selection use a convex optimization formulation that imposes a 1-norm penalty on the elements of the inverse covariance matrix.  This approach can be extended to group penalties for networks with certain structured sparsity patterns. Another interesting extension calculates the effect of latent variables on the network topology by combining penalty functions that encourage sparsity and low rank. The talk will present an overview of these approaches and corresponding algorithms.  We will discuss in some detail first-order methods based on splitting techniques, such as the Alternating Direction Method of Multipliers (ADMM) and its variants. The methods will be illustrated with results for environmental (air quality) and functional magnetic resonance imaging (fMRI) data sets.
"Energy-Efficient Spectrum Sharing Among Decentralized Users with Binary Feedback"
Yuanzhang Xiao and Mihaela van der Schaar 
Abstract: We develop a novel design framework for energy-efficient spectrum sharing, in which autonomous users aim to minimize their transmit power levels subject to their minimum throughput requirements. When the number of users is large or the multi-user interference is strong, the optimal spectrum sharing should be in a TDMA fashion. The proposed spectrum sharing policy enables autonomous users, using (possibly erroneous) binary feedback, to achieve TDMA in a decentralized manner. Compared to state-of-the-art policies, the proposed policy can triple the achievable average throughput, and achieve an energy saving of up to 80% under the same throughput requirements. 

Session 14. Next-Generation DSP Architectures 

"Signal-processing Techniques for Wideband Data Converters" 
Abhishek Ghosh and Sudhakar Pamarti
Abstract: This talk will focus on signal conditioning techniques for time-interleaved A/D converters. Time-interleaving is an effective way to enhance ADC bandwidths without compromising their dynamic range. However, time-interleaved ADCs are plagued by mismatches between the constituent channels in terms of offset, gain, sampling-instants, bandwidths etc, which severely curtail the effective dynamic range attained in such ADCs. A novel adaptive signal-conditioning technique is proposed to correct for the channel-mismatch errors, thus reclaiming about 30dB of the overall dynamic range. The proposed technique is significantly low-power compared to competitive techniques in similar CMOS technologies evident from simulated digitization of signals having bandwidths in the excess of 150MHz, up to a 10.5 bits resolution consuming about 5mW of power. 
"Flexible DSP Architecture for Next-Generation Software-Defined Radios" 
Fang-Li Yuan and Dejan Markovic
Abstract: Future software-defined radios require fast switching between communication functionalities, algorithms and standards depending on the channel condition and available spectrum resources. A multi-standard design is either energy-inefficient by using programmable DSPs, or inflexible to tackle any future adjustments with dedicated ASICs. Our research focuses on exploring a new architecture that allows the coexistence of flexibility and efficiency. Three key design techniques regarding granularity, connectivity, and controllability are discussed. The proposed multi-core flexible DSP is able to bridge the performance gap between DSPs and ASICs, reaching >5 GOPS/mW under 40nm CMOS technology.
"Sparse Signal Decoder for Compressive Sensing Applications" 
Fengbo Ren and Dejan Markovic
Abstract: In recent years, compressive sensing (CS) has attracted extensive attentions. The CS framework offers a compressive sampling scheme that directly encodes the sparse representation of a signal into very few measurements, allowing for data sampling with minimum redundancy at sub-Nyquist rates. This leads to great application values for mobile applications. However, the signal decoding involves a complicated optimization problem. So far, rare promising results have been demonstrated in hardware implementation. In this work, we are designing a flexible and efficient sparse signal decoder in VLSI implementation, aiming to provide a viable solution to the real-time signal processing for CS applications.
"Integrating Dynamic Economic Optimization and Model Predictive Control for Optimal Operation of Nonlinear Process Systems" 
Matthew Ellis1 and Panagiotis D. Christofides1,2
1Dept of Chemical & Biomolecular Engr; 2Dept of Electrical Engr
Abstract: Economic optimization of chemical processes has traditionally been addressed through real-time optimization (RTO). In general, an RTO system carries out economic optimization and computes optimal process operation steady-states using steady-state process models. In this work, we propose a conceptual framework for integrating dynamic economic optimization and model predictive control (MPC) for optimal operation of nonlinear process systems. First, we introduce the proposed two-layer integrated framework. The upper layer, consisting of an Economic MPC (EMPC) system that operates outside of the feedback control loop, computes economically optimal time-varying operating trajectories for the process by optimizing a time-dependent economic cost function over a finite prediction horizon subject to a nonlinear dynamic process model. The lower feedback control layer may utilize conventional MPC schemes or even classical control to compute feedback control actions that force the process state to track the time-varying operating trajectories computed by the upper layer EMPC. Such a framework takes advantage of the EMPC ability to compute optimal process time-varying operating policies using a dynamic process model instead of steady-state models, and the incorporation of suitable constraints on the EMPC allows calculating operating process state trajectories that can be tracked by the control layer. Second, we prove practical closed-loop stability including an explicit characterization of the closed-loop stability region. Finally, we demonstrate through extensive simulations using a chemical process model that the proposed framework can both (1) achieve stability and (2) lead to improved economic closed-loop performance compared to real-time optimization (RTO) systems using steady-state models.
Document Actions