Zhi DIng - US grants

Blind adaptive equalizers are important elements in high data rate digital communication systems to remove linear distortions incurred by nonideal and bandlimited channels. By removing linear channel distortions without the aid of a training sequence, blind equalization makes it unnecessary for the transmitter to interrupt the normal data transmission in order to generate a training signal for the sake of a newly tuned receiver. Despite the plethora of existing blind equalization schemes, few exhibit the desired global convergence property. Many existing blind equalizers are prone to undesirable local convergence corresponding to little or no intersymbol interference removal, as shown in the previous works by the principal investigator and others. In addition, the convergence rates of many of the existing algorithms are very slow. A principal objective of this research is to improve the performance of blind equalization systems by designing new and better blind equalization algorithms. Various linear and nonlinear constraints will be utilized in developing, analyzing and testing fast and globally convergent equalization algorithms. The cyclostationary nature of the PAM/QAM modulated channel output will be exploited to generate innovative designs of blind equalizers. In addition, various stochastic dynamic systems analysis tools (such as probability and stochastic processes, spectral analysis, averaging and stability theory) will continue to be exploited in this study, as it has been in many previous adaptive system studies. However, unlike most of the previous studies, cyclostationary signal analysis will also be utilized in the design of blind equalizers based on the cyclostationarity of the channel output. The proposed research is aimed at expanding the utility of the concept of blind adaptive equalization and applications of broadcast digital communication systems, and is expected to have significant direct and indirect impact on the application of adaptive systems in areas such as adaptive system control, robotics, array processing, mechanical and aerospace engineering, and geophysical signal processing.

0121469
Ding

The imminent commercial deployment of the 3rd generation 3rd wireless communications systems has placed the development of future high rate wireless internet and mobile computing at the top of wireless R&D agenda. Wireless channels are known to be unreliable and volatile. To develop even faster wireless data networks that can serve as the essential last link for both nomadic and highly mobile users, wireless system design may require fundamental changes in approach. To this end, it has been widely recognized that quality-of-serve (QoS) awareness should be introduced to lower network layers through a unified design. Even within the physical layer itself, separate function blocks such as transmit diversity, equalization, and coding must be further integrated to better exploit the available channel bandwidth, antenna diversity, and channel coding in order to compensate for unknown and time-varying channel distortions as well as co-channel interferences. This proposal presents a novel framework for the design integration and VLSI implementation of ARQ based wireless data networks and highlights a number of critical open research issues for investigation along with many highly promising solutions.

The proposed project centers on a unified approach to the design integration of ARQ, transmit/receiver diversity, equalization, and channel coding in broadband wireless communication systems. It represents a clear departure from the current transceiver (physical layer) design approach that separately determines and optimizes each individual functionality such as channel coding, coded modulation, and channel equalization. Central to the PIs proposal is the idea to exploit the use of QoS driven hybrid ARQ at both the transmitter and the receiver. This idea represents an important step of design integration based on the network use of hybrid ARQ. The PIs proposal focuses on a specific mechanism facilitated by an ARQ feedback channel to achieve design coordination of physical layer functions of channel equalization/pre-compensation, transmit antenna diversity, (turbo) coding, amid hybrid ARQ. Their proposed approach is very general and is not limited to a specific modulation or network standard. In particular, their research objectives amid tasks include

o Joint optimization of turbo coding, turbo equalization, amid multiuser detection through code awareness based on temporal diversity introduced by hybrid ARQ;

o Turbo equalization amid multiuser detection algorithms for transmit diversity with variable rate codes that are exploited by the transmitter amid demodulator;

o Broadband spatial turbo equalization combined with H-ARQ diversity utilizing codes of different levels of error protection as well as iteratively decodable multilevel codes;

o New type II hybrid ARQ based on integration of parity retransmission, concatenation, Reed-Solomon (RS) outer code, low-density parity check (LDPC) inner code, arid turbo coding.

o VLSI design amid implementation of high speed decoders for integrated equalization, LDPC, and RS turbo decoding.

o Low complexity two-stage turbo decoding of Reed-Solomon codes through their binary decomposition into binary component codes with relatively small state complexity;

Rather than pursuing incremental advances separately in diversity combining, H-ARQ, equalization, and coding, this project presents a new and specific network integration approach that will have a strong impact on the design of future wireless data networks. The VLSI design amid implementation will represent the latest technology for broadband high speed communication systems.

This research project investigates the design, analysis, and implementation of resource efficient integrative transceivers and retransmission diversity in broadband multi-input-multi-output (MIMO) wireless communications. The development and exploitation of MIMO technologies and Automatic Repeat reQuest (ARQ) protocols have been popular subjects of wireless research. Hybrid ARQ is effective as protection against packet error in wireless communications, while MIMO transceivers have demonstrated significant performance gains at the wireless physical layer. However, traditional approaches treat MIMO schemes and ARQ as independent mechanisms in wireless networks. The design integration of hybrid ARQ protocols with MIMO transceivers has received scant coverage and has not been well utilized. As more and more mainstream products begin to adopt MIMO technologies in wireless LAN and other wireless systems, there is an urgent need to exploit and achieve the full potential benefit offered by integrating wireless ARQ and MIMO designs. This research project investigates the efficient integration of ARQ with broadband MIMO physical layer.

Simultaneously considering both designs of ARQ protocols and MIMO transceivers, the investigations of this project are systematic, integrative, and broadly applicable. Unlike in traditional wireless designs where ARQ and MIMO technology are treated separately, this integrative design approach opens a new door to developing future advanced wireless systems that are power and bandwidth efficient. While the advantages of cross-layer optimization are well known, the integrative design of ARQ protocols and their corresponding MIMO transceivers provides a concrete and tangible approach to cross-layer network design. Focusing on bandwidth and power efficiency, the project goal is to optimize and adapt ARQ (Automatic Repeat reQuest) protocols for MIMO wireless systems to improve performance and scalability. The research presents new capacity analysis and transceiver optimization of progressive MIMO precoding to better exploit the ARQ retransmission diversity. The investigators develop design of bandwidth and power efficient transceiver technologies in full integration with new scalable ARQ designs to significantly improve the performance of end user applications. The results of this project provide new formulation of wireless MIMO channel estimation algorithms particularly tailored for bandwidth efficient and scalable hybrid ARQ protocols in future broadband wireless communication systems.

Abstract
Localization involves the estimation of the precise location of an object based on various forms of relative position information available of the object. Source and sensor localization is a fundamental capability broadly useful in a number of emerging applications. For example a network of sensors deployed to combat bioterrorism, must not only detect the presence of a potential threat, but also must pinpoint the source of the threat. Similarly, in pervasive computing, locating an errant mobile user permits the computer network to identify the most appropriate server with matching capabilities for the user. Likewise, in sensor networks individual sensors must know their own positions, so as to route packets, detect faults, and sense and record events. There is also an emerging multibillion dollar wireless localization industry. This research will address issues that hold the key to fast efficient localization.
The investigators will adopt a three pronged approach. First, various optimum estimates will be investigated under a variety of practical signal models. These include maximum likelihood and minimum variance estimates. Theoretical performance limits will be determined. Secondly, the investigators will generalize and analyze various algorithms that obtain these estimates efficiently with low complexity. Specifically, the investigators will study optimal localization involving minimization of non-convex cost functions that admit multiple local minima. To overcome the problem of multiple local minima, the investigators will develop a relaxation framework based on convex optimization to obtain fast near optimal solutions. Finally, the proper placement of wireless sensors and anchors impacts both the accuracy and the complexity with which localization is performed. Thus, the investigators will study optimum anchor placement to aid both these attributes. These investigations are critical to the understanding of the theoretical foundation of wireless source localization, and will fundamentally impact its broad applications.

The project will investigate a new ultra-high capacity communication technology that can scale beyond terabit per second capacity, withstand physical layer impairments, and flexibly support diverse data flows. The project combines new and scalable optical arbitrary waveform generation techniques together with microwave communications, signal processing, and coding techniques. The resulting terascale system effectively accommodates high and low data flow traffic and mitigates physical layer impairments. The project will also investigate CMOS-compatible silicon based optical integration, and will pursue future integration of electronic and photonic functions to realize more compact and agile functions on a chip scale system.

The intellectual merit of this research is in the theoretical and experimental investigations of hardware and software, and in studying photonic and electronic solutions for future terascale communication systems with resiliency, flexibility, and scalability. Furthermore, the project will also pursue significant enhancement in signal processing capability by combining best opportunities in both optics and electronics. The results of this project will provide a new insight into optical-electronic communication technologies, hardware-software codesign, silicon photonic integrated circuits, and future prospects for resulting communication, computing, and information technology of the future.

The broader impacts of this research are as follows: (1) transforming the way of design, implementation, and operation of healthcare, data center networking, and cloud computing, (2) making available compact communication tools fabricated by CMOS foundry to provide rapid and widespread benefits to the society, (3) bringing new communication theory and algorithm crosscutting optics, electronics, and computing framework, (4) developing new courses to disseminate the up-to-date knowledge on communications and optoelectronics, (5) introducing opportunity for underrepresented and high-school students to be involved in state-of-the-art technology research and education, and (6) stimulating new opportunities towards realizing intelligent terascale communication systems.

This NSF collaboration project investigates the design and optimization of dynamic resource allocation and interference management strategies for spectrally efficient wireless heterogeneous networks that provide multi-level coverage and services. This research project is motivated by the recent advances in 4G wireless systems on the deployment and standardization of heterogeneous networks (HetNet). Such networks provide services to mobile subscribers of different priorities and dynamic quality needs. As more and more advanced physical layer schemes and medium access protocols are being integrated into wireless standards, future performance gain and progress in wireless networks have to rely more on intelligent resource allocation and interference management strategies that are dynamical and are adaptively responsive to location-and-time specific environment. In this project, the research team will address critical deployment issues that arise in HetNet by focusing on the development of distributed and effective mechanisms for resource allocation and interference management in order to facilitate low complexity and decentralized network operation in heterogeneous environments. The project methodology is based on novel optimization frameworks for interference control and suppression in HetNet. The research goal is to develop robust and reliable solutions for practical implementations of HetNet. The project results will facilitate novel technological directions that transcend multiple networks and multiple network layers. In particular, the results will assist the near term deployment of wireless HetNet, including the broad application of femtocell deployment.

The technical impacts of this international collaboration project are broad both domestically and internationally. Results from successful execution of the project are expected to significantly impact the design, deployment, and operation of future wireless networks. The plan to disseminating research findings at quality journals and technical conferences will contribute directly to the wireless industry by providing critical information on interference control and management for high spectral efficiency and user satisfaction. The project outcomes can also establish new research directions for the international telecommunications community. The project activities will lead to new analytical tools and discoveries that can impact other science and engineering research fields. The project will further contribute to the training of highly qualified personnel for the hi-tech industry.

This work develops a novel receiver methodology to integrate signal detection and forward
error correction in multiple-input-multiple-output (MIMO) communications and various diversity
transmissions. Moving beyond traditional approaches relying on belief-propagation, this
investigation into the rather classic open problem of integrated signal detection and decoding
involves novel joint optimization formulations that incorporate binary field parity constraints
imposed by the low-density parity check within maximum likelihood detection frameworks for
unified optimization. This novel framework is general and encompasses a number of practical
transmission models, including distributed antennas, opportunistic cooperative networking,
and signal retransmission as well as their integrations. This project has broad impacts on
engineering, education, and society. Its success can lead to new research directions, new
tools, and results to help advance other science and engineering fields.

Focusing on multicarrier MIMO signal reception, the investigators will develop and optimize
integrated receivers for important wireless network diversities including distributed
transmissions, cooperative MIMO, and hybrid-ARQ retransmission systems. The new design
methodology emphasizes integration of multiple constraints from incompatible fields. By
reformulating and relaxing joint detection and decoding problems into convex optimization, the
investigators will design high performance receivers that are efficient and are robust to various
forms of uncertainties in channel state information. Such a novel approach to receiver design
integration represents a fundamental and practical design paradigm that can fully utilize
various a priori signaling and code constraints for joint detection and decoding against channel
distortions and other non-idealities to achieve high performance, efficiency, and reliability. The
research findings can contribute practically to improvement of future wireless services and
broaden their applications in many practical fields where quality, efficiency, and distributivity
concerns are paramount.

The dynamic and uneven wireless network traffic load at different time instants and geographical locations has led to substantial underutilization of some spectrum bands while severely crowding others. This project investigates a number of fundamental and challenging technical issues that arise from broadband spectrum trading for achieving superior technical, economic, and social values of spectrum use. The project is an interdisciplinary research effort across mathematics and wireless network technologies. With respect to wireless technologies, this research project outcomes can significantly improve spectrum efficiency and user experience, while benefiting many real-life needs, such as public safety, telemedicine, and social services. On mathematics, the research effort will lead to the formulation of more interesting problems with real world applications and the discovery of new tools for solving such problems.

As a comprehensive investigation to overcome technical challenges that arise from broadband spectrum trading, the project tasks focus on three inter-related key research directions: 1) new graph theory problems, 2) new graph-based resource allocation and utility optimization, and 3) physical layer techniques and utility design. Specifically, the project considers new problems on judicious partitions of graphs and new problems on optimal disjoint paths in weighted graphs using minimum-edge-weight path utility. New solutions and tools developed for such problems can then be applied and generalized for solving various resource allocation and utility optimization problems in broadband spectrum trading. Furthermore, the project addresses new fundamental physical layer issues that arise from spectrum trading. These implementation issues include broadband channel estimation, dynamic pilot placement, interference limited pilot power control, and low complexity broadband spectrum sensing. To establish utility functional curves and to develop means for modeling parametric effects in fine-granularity for more effective broadband spectrum trading, the project applies group-theory-based methodologies to design and carry out detailed tests and analysis in order to fine-tune effective utility functions that incorporate user experience and satisfaction. The research results are also expected to lead to strong social impacts and to provide better technical insights and effective guidelines on governmental regulatory policymaking and technological development regarding broadband spectrum trading for wireless communications.

This project seeks an innovative approach to the design of very high-bandwidth and ubiquitous wireless networks exploiting the unique properties of RF-photonic signal processing and integrated photonics for emerging next-generation and high-bandwidth 5G technologies. The flexibility offered by the photonic and electronic system co-design has the potential to provide a 100-fold increase in bandwidth available for the mobile users, compared to what currently available with current technology. This project will exploit key enabling physical layer technologies, such as dynamic optical waveform generation and measurement (OAWG and OAWM) on silicon photonics (SiP), together with SiP lattice filters (SiPhaser) and photonic mixers. The combination of these technologies will provide a unique and efficient front-haul architecture to perform analog massive-MIMO beamforming at mm-wave frequencies without requiring any complex high-speed RF circuitry. The following topics will be investigated: (1) RF-Optical Networking architecture design and performance studies; (2) DSP, coding, and mmWave-MIMO algorithms development; (3) simulation studies of analog RF-optical beamforming by SiPhaser filters; (4) Proof-of-principle demonstration of SDM MIMO mmWave with RF-photonic processing.

Future Internet applications will exponentially increase the demand for ubiquity, mobility, and bandwidth through diverse platforms, in particular, rapidly expanding cloud data center infrastructures. Today's wireless networks are typically limited to 1~100 Mb/s connectivity with limited end-to-end throughput, latency, and reliability. While fiber optic networks provide 1~100 Tb/s capacity on each single-mode-fiber over thousands of kilometer distances, they are limited to wide area and metro networks, not readily accessible by mobile users. This project seeks to improve our nation's cyberinfrastructure by making additional bandwidth available to citizens with mobile devices. The project will also provide an exciting opportunity to train students in design, developing, and testing algorithms and physical layer technologies on a futuristic networking platform.

This research project addresses some critical and emerging challenges in future generations of wireless communication networks to serve the need of the widespread Internet of Things (IoT) applications. Unlike existing cellular networks, IoT applications are expected to connect a massive number of smart IoT devices under severe bandwidth constraint. This massive scale and need of IoT wireless devices, along with many other rising wireless applications such as vehicular to everything (V2X) strongly motivates highly effective utilization of network capacity. To substantially expand network capacity, non-orthogonal multiple access (NOMA) is a cutting-edge technology that integrates the concepts of superposition coding at transmitter and successive interference cancelation (SIC) at receiver, respectively. NOMA has attracted much attention for its high spectral efficiency. However, the NOMA throughput gain comes at the price of having to overcome substantial co-channel interferences (CCI) among wireless network nodes. Despite the widespread and idealized receiver assumption of perfect interference cancellation, practical success of NOMA receivers depends critically on the successful detection and decoding of signals in the presence of substantial CCI. This research project aims to develop highly effective and practical joint detection and decoding receivers to deliver the much needed network performance for successful deployment of NOMA in future 5G and IoT wireless services. Specifically, the project develops optimized design of joint detection and error correction receivers by leveraging the knowledge of user forward error correction codes to substantially improve receiver performance against CCI under channel uncertainties. The research findings can contribute importantly to the service improvement of high speed wireless networks and to broadening their applications in many practical fields where quality, efficiency, and service decentralization are paramount. The success of the project can lead to new system designs, new tools, and results that can impact other science and engineering fields.

The technical focus of this project is to develop a new receiver design methodology that unifies signal detection and forward error correction in multiple-input-multiple-output (MIMO) wireless communications and diversity networks against poor wireless channel conditions. Unlike traditional approaches, this new investigation into the rather classic but open problem centers on novel optimization formulations that can incorporate Galois field codeword constraints imposed by the forward error correction (FEC) codes within the maximum likelihood detection principle for unified receiver optimization. This novel framework is general and encompasses many wireless models, including distributed MIMO, opportunistic cooperative networking, and retransmission diversities as well as their integrations. This innovative direction emphasizes critical integration of multiple constraints from incompatible fields through effective constraint relaxation and novel objective function formulation. Reformulating the optimization of joint detection and decoding problems into convex optimization, the proposed approach to receiver design integration represents a fundamental and practical design paradigm that can fully leverage various practical signaling and code constraints for joint detection and decoding against channel and other non-idealities to achieve high performance, efficiency, and reliability. To confront practical challenges of channel estimation errors and fast channel fading in wireless networks, the project team shall develop fast, effective, reliable, and robust algorithms for coded MIMO transmissions under strong co-channel interferences subject to different practical limitations and network configurations, with respect to complexity and performance tradeoffs. The research findings are expected to have significant broader impact on a wide range of wireless applications including high speed cellular, IoT, and V2X services.

The Internet-of-Things (IoT) has recently emerged as a powerful new paradigm for future generations of wireless networks. In IoT, a wide variety of devices, such as body sensors, implants, animal biochip transponders, electric clams in coastal waters, vehicular sensors, sensors for environmental/food/pathogen monitoring, and devices in disaster relief operations, are connected to the Internet via wireless interfaces. Connecting this myriad of mobile devices to the Internet could potentially lead to a broad range of innovative network applications. However, unique technical challenges for IoT, such as massive connectivity, security vulnerability and energy sustainability, among others, need to be addressed before such potentials can be fully fulfilled.

This project aims to develop a robust and secure framework for wireless function computing to specifically address challenges in IoT applications. The enabling characteristic of this framework is that in many IoT applications, instead of recovering the full data collected by various devices and transmitted over the network, the goal of the networked functional computing is to assist certain decision making, for which the decision makers only need to compute a function of these distributed data. The main idea of this project is to develop secure and spectrum/energy efficient protocols that enable the decision maker to compute functions of interest without first recovering the full data from sensing devices. Thus, instead of acting as mere data pipes, wireless links become an integral part of the smart decision process in IoT applications. Successful execution of this project will make substantial contributions to both practical applications and theory development. On the application level, this project has the potential to substantially improve the spectrum efficiency, energy efficiency and robustness to active attacks for future IoT networks. On the theory side, this project will develop low-complexity efficient and robust function computing algorithms that enable IoT applications to make timely and accurate decisions.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

The recent wave of technological advances in machine learning and artificial intelligence has led to widespread applications and public awareness. At the same time, the rapid growth of high-speed wireless network services presents an opportunity for future distributed learning involving a vast number of smart IoT devices. This project targets several technical challenges posed by the limited reliability of wireless connections and computational constraints of the edge nodes in distributed learning systems. Overcoming these challenges is vital to the plethora of computation, communication, and coordination tasks required by distributed machine learning at the network edge. Centered on developing innovative edge learning algorithms over wireless MAC channels under the constraints of computing, power, and bandwidth, this project can significantly impact wireless edge learning in a variety of IoT applications, ranging from transportation, safety, and agriculture, to energy efficiency, e-health, and smart infrastructure. The broader impact of this research will also come through many educational opportunities by providing opportunities in STEM to K-12, women, and underrepresented minority students.


This collaborative project will develop an innovative network architecture for distributed learning over wireless multi-access channels. Specifically, the PIs will take a principled approach to develop an integrated wireless edge learning framework, using both gradient-based methods and also very recent advances in gradient-free, zero-order optimization, while taking into account the constraints in computing, power and bandwidth therein, in a holistic manner. The developed methods will be also extended to the setting of distributed online learning and reinforcement learning under wireless MAC. The PIs will focus on optimizing communication-efficient gradient sparsification based local updates that are communicated within the wireless network under bandwidth constraints; and each sender intelligently carries out transmission power allocation based on learning gradient and channel conditions. One important objective is to develop a novel learning-based framework for efficient wireless channel estimation and update to enable effective power control and learning. The project will devise edge learning algorithms that are robust against wireless channel uncertainty. The team of PIs shall comprehensively investigate the impact of the wireless bandwidth and power constraint on both the accuracy and convergence speed of edge learning algorithms.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.