Seminar Schedule

Seminar Home   Hide Abstracts   PDF   Plain Text
Chronological By Speaker
 
Title A Horizontal Time Decomposition Strategy for Multi-Interval Scheduling of Power Systems
Speaker Farnaz Safdarian
Ph.D. Candidate, Division Electrical & Computer Engineering
Louisiana State University
Abstract

In this presentation, a horizontal time decomposition and distributed coordination strategy will be introduced to reduce the computation time of power system multi-interval scheduling problems. The considered scheduling horizon is decomposed into multiple smaller sub-horizons. The concept of overlapping time intervals is introduced to model ramp constraints of generating units for transition from one sub-horizon to another sub-horizon. A sub-horizon includes n internal intervals and one or two overlapping time intervals that interconnect consecutive sub-horizons. A local optimization problem is formulated for each sub-horizon with respect to internal and overlapping intervals variables/constraints. The overlapping intervals allow modeling intertemporal constraints between the consecutive sub-horizons in a distributed fashion. To coordinate the subproblems and find the optimal solution for the whole operation horizon distributedly, accelerated auxiliary problem principle is developed. Furthermore, we present an initialization strategy to enhance the convergence performance of the coordination strategy.

Bio

Farnaz Safdarian is currently a Ph.D. student in the Division of Electrical and Computer Engineering at Louisiana State University. She has received her B.S. and M.S. degrees from Shahid Beheshti University and Amirkabir University of Technology (Tehran Polytechnic), Iran, in 2011 and 2014, respectively. Her research interests include power systems operation and planning, decentralized/distributed optimization, smart grids, renewable energy, HVDC systems, and power electronics.

When Tuesday, 22 January 2019, 12:30 - 13:30
Where Room 3250H Patrick F. Taylor Hall
More Hide Abstracts.   Announcement (PDF)
 
Title Distributed Energy Management in Smart Grids: From Theory to Practice
Speaker Javad Mohammadi
Carnegie Mellon University
Abstract

It is expected that the power grid of the future differs from the current system by the increased integration of distributed generation, distributed storage, demand response, power electronics, and communications and sensing technologies. The consequence is that the physical structure of the system becomes significantly more distributed. The existing centralized control structure is not suitable any more to operate such a highly distributed system. In this talk I will introduce a promising solution to a class of energy management problems in power systems with a high penetration of distributed resources. This class includes optimal dispatch problems such as optimal power flow, security constrained optimal dispatch, optimal power flow control. The proposed distributed framework is based on iteratively solving in a distributed fashion the first order optimality conditions associated with the optimization formulations. A multi-agent viewpoint of the power system is adopted, in which at each iteration, every network agent updates a few local variables through simple computations, and exchanges information with neighboring agents. The proposed distributed solution is based on the consensus+innovations framework, in which the consensus term enforces agreement among agents while the innovations updates ensure that local constraints are satisfied. Finally, I will present practical examples to showcase that the proposed fully distributed algorithm not only handles the computational complexity of the problem, but also provides a more practical solution for these problems in the emerging smart grid environment.

Bio

Javad Mohammadi a special faculty in Electrical and Computer Engineering at Carnegie Mellon University. Prior to this, he was with the Electrical and Computer Engineering Department at CMU as a research scientist. He received my Ph.D. in Electrical and Computer Engineering from CMU in 2016. My research interests include energy system optimization and control, distributed computations, smart grid, and electrified transportation systems.

When Friday, 8 February 2019, 13:30 - 14:30
Where Room 1212 Patrick F. Taylor Hall
More Hide Abstracts.   Announcement (PDF)
 
Title Operational Equilibria of Electric and Natural Gas Systems with Limited Information Interchange
Speaker Antonio Conejo
Ohio State University
Abstract

Electric power and natural gas systems are typically operated independently. However, their operations are interrelated due to the proliferation of natural gas-fired generating units. We analyze the independent but interrelated day-ahead operation of the two systems. We use a direct approach to identify operational equilibria involving these two systems, in which the optimality conditions of both electric power and natural gas operational models are gathered and solved jointly. We characterize the equilibria that are obtained under different levels of temporal and spatial granularity in conveying information between the two system operators. Numerical results from the Belgian system are used to examine the impacts of different levels of information interchange on prices and operational cost and decisions in the two systems.

Bio

Antonio J. Conejo, professor at The Ohio State University, OH, US, received the B.S from Univ. P. Comillas, Spain, the M.S. from MIT, US and the Ph.D. from the Royal Institute of Technology, Sweden. He has published over 165 papers in SCI journals and is the author or co-author of books published by Springer, John Wiley, McGraw-Hill and CRC. He has been the principal investigator of many research projects financed by public agencies and the power industry and has supervised 19 PhD theses. He is an IEEE Fellow.

When Monday, 11 February 2019, 11:30 - 12:30
Where Room 1202 Patrick F. Taylor Hall
More Hide Abstracts.   Announcement (PDF)
 
Title Quality of Time: A New Perspective to Design Cyber-Physical Systems
Speaker Fatima Anwar
University of California, Los Angeles
Abstract

Unprecedented Cyber-Physical Systems (CPS) applications such as health care, connected vehicles and augmented/virtual reality are revolutionizing smart spaces. These applications span the cloud and edge devices with a critical dependence on temporal use cases. As such, cloud services are expected to provide timely responses and schedulable demands, while edge devices are required to synchronize observations and choreograph actions across distributed entities. The goal of my research is to design new systems that enable time awareness and meet consistency, causality and scheduling demands of underlying CPS applications running on commodity platforms. In particular, I design trustworthy systems centered around extensible time abstractions in the presence of timing variations and vulnerabilities.

In this talk, I will first discuss the challenges faced by time-aware applications. I will then motivate and present the necessary timing abstractions that treat time as a controllable operating system primitive while taking into account the uncertainty arising from hardware and network variations. Further, I will discuss timing vulnerabilities in trusted execution technologies and network security mechanisms; and present my design of secure global clocks. While my abstractions and system designs can be applied to many CPS applications, my talk will focus on autonomous driving use cases.

Bio

Fatima Anwar is a Ph.D. candidate in the Electrical & Computer Engineering department at UCLA. Her research interests lie in the intersection of system design, security, and quality of time in distributed Cyber-Physical Systems. Specifically, she designs trustworthy systems around abstractions to provide key services to the Internet of Things applications running on commodity platforms and operating systems. Earlier, she used to work at Samsung Electronics on the Smart Health project (SHealth) and developed a sensor service framework for mobile devices. She was Qualcomm Innovation Fellowship finalist in 2018 and was Anita Borg scholar in 2017. Fatima is also committed to broadening participation and volunteers for Los Angeles Computing Circle (LACC) and Engineering day for Girls at UCLA.

When Thursday, 28 March 2019, 9:15 - 10:15
Where Room 3107 Patrick F. Taylor Hall
More Hide Abstracts.   Announcement (PDF)
 
Title Trying to Keep it Real: 25 Years of Trying to Get the Stuff I Learned in Grad School to Work on Mechatronic Systems
Speaker Daniel Abramovitch
Agilent Technologies
Abstract

This talk is about the difficulty of making well known and widely accepted advanced textbook control techniques work in an industrial environment, particularly with mechatronic systems that have large numbers of flexible modes. I will go through the methods that fail if done the standard way and the adjustments I have learned to make over the years which get a lot of them to work. I will also go over the methods that seem to work robustly and without much thought in the industrial environment, explaining why they do work. Finally, I will try to show that understanding the differences and commonalities in these two world views allows us to use the principles of one to improve the other.

Bio

Danny Abramovitch earned degrees in Electrical Engineering from Clemson (BS) and Stanford (MS and Ph.D.), doing his doctoral work under the direction of Gene Franklin. Upon graduation, and after a brief stay at Ford Aerospace, he accepted a job at Hewlett-Packard Labs, working on control issues for optical and magnetic disk drives for 11 1/2 years. He moved to Agilent Laboratories shortly after the spin off from Hewlett-Packard, where he has spent 19 years working on test and measurement systems. He is currently in Agilent's Mass Spectrometry Division working on improved real-time computational architectures for Agilent's mass spectrometers.

Danny is a Senior Member of the IEEE and was Vice Chair for Industry and Applications for the 2004 American Control Conference (ACC) in Boston. He was Vice Chair for Workshops at the 2006 ACC in Minneapolis, for Special Sessions at the 2007 ACC in New York, and for Industry and Applications for the 2009 ACC in St. Louis. He was Program Chair for the 2013 ACC and is General Chair of the recent 2016 ACC in Boston. He has helped organize conference tutorial sessions on topics as varied as disk drives, atomic force microscopes, phase-locked loops, laser interferometry, and how business models and mechanics affect control design. He served as the Chair of the IEEE CSS History Committee from 2001 to 2010. Danny is credited with the original idea for the clocking mechanism behind the DVD+RW optical disk format and is co-inventor on the fundamental patent. He was on the team that prototyped Agilent's first 40Gbps Bit Error Rate Tester (BERT) and was able to cite a Douglas Adams book in one of his patents relating to that device. Along with his co-author, Gene Franklin, he was awarded the 2003 IEEE Control Systems Magazine Outstanding Paper Award. His favorite paper remains the one prompted by a question from his then 3-year-old son, which showed that the outrigger was a feedback mechanism that predated the water clock by at least a 1000 years. He was a Keynote Lecturer at the 2015 Multi-Conference on Decision and Control in Sydney, Australia. His recent work for Agilent was on high speed atomic force microscopes and high precision interferometers. His current work involves improving the real-time control, data collection, and signal processing chain on Agilent's Mass Spectrometers. He is part of the team that introduced the multi-award winning Ultivo Tandem Quad Mass Spectrometer in 2017. He is the holder of over 20 patents and has published over 50 reviewed technical papers.

When Wednesday, 3 April 2019, 10:30 - 11:30
Where Room 3316E Patrick F. Taylor Hall
More Hide Abstracts.   Announcement (PDF)
 
Title Partition of Random Items: Tradeoff between Binning Utility, Meta Information Leakage and Hypotheses Distinguishability
Speaker Farhang Bayat
Louisiana State University
Abstract

In this talk, we will present a novel framework to understand the tradeoff between binning utility, meta information leakage and hypotheses distinguishability. More specifically, under the proposed framework, we formulate two constrained optimization problems. In the first problem, the goal is to maximize the binning utility while restraining a certain level of information leakage. In the second problem, the goal is to maximize the binning utility while maintaining a lower bound for the measure of distinguishability between two hypotheses and an upper bound for the level of information leakage. Both problems are NP-hard by nature because we are seeking an optimal allocation of M random items into N bins. Such problems (where optimization is carried out over a diverse but yet dependent series of sets) are formally known as multi-agent multi-variate optimization problems. We develop suboptimal solutions to exploit potential sub-modular structures of the problems based upon sufficient conditions found on the payoff functions in the underlying objective functions. Further numerical results are presented to demonstrate the effectiveness of the proposed sub-modular function based algorithms.

Bio

Farhang Bayat received the B.Sc. and M.Sc. degrees in electrical engineering from Amirkabir University of Technology, Tehran, Iran, in 2012 and 2014, respectively and is currently pursuing the Ph.D. degree in the Division of ECE of the School of EECS at Louisiana State University, Baton Rouge. His research interests include information theory and its applications in privacy protection in communication networks and graphical models.

When Thursday, 4 April 2019, 12:30 - 13:30
Where Room 3285 Patrick F. Taylor Hall
More Hide Abstracts.   Announcement (PDF)
 
Title How Entergy's Transmission Operation Planning Team Works: Introduction and Job Opportunities
Speaker Michelle P. Bourg and Paul Simoneaux, Jr.
Entergy
Abstract
Bio

Ms. Bourg has over sixteen years of utility experience, including roles in Entergy's utility, human resources, transmission and gas businesses. She was recently appointed to her current role of vice president, transmission operations. In this capacity, she is responsible for providing strategic and executive leadership to approximately 130 transmission operations management and support staff, including engineers, real-time system operators, trainers, and other professionals. Her organization is responsible for safely and reliably operating the Entergy electric transmission system while maintaining compliance with all applicable federal regulations.

Ms. Bourg began her career as an engineer in the transmission operational planning group and later managed the workgroup through several significant changes, including the response to increased compliance oversight and various regulatory directives. She later served as the director of utility performance metrics where she was responsible for the deployment of tools to assist employees at all levels monitor and improve performance results. She most recently led Entergy Louisiana, LLC and Entergy New Orleans, LLC's gas distribution business where she was responsible for the safe, reliable and efficient delivery of natural gas to customers in New Orleans and Baton Rouge.

She has a bachelor's degree in electrical engineering from Louisiana State University and a master of business administration from Tulane University. She is a registered professional engineer in the state of Louisiana. She and her husband, Jonathan, have three young children – Coleman (8), Caroline (5) and Claire (2). In her spare time, she enjoys spending quality time with her children and family.

Paul Simoneaux, Jr. was named manager of transmission operational planning in June 2014. Immediately prior to being named to this position, he served as the supervisor for transmission asset management.

The manager of transmission operational planning manages the workgroup responsible for providing technical support to real-time operations staff at the Transmission Control Centers (TCCs).  This support is accomplished through the preparation of reliability power flow models and corresponding real-time, next-day and forward looking analyses, coordination of planned transmission outages, voltage stability analysis, development of operating guides and local area security analysis.  The Manager, Transmission Operational Planning provides input to regulatory/compliance filings in support of new business practices for Transmission Operations and ensures that the workgroup processes are in compliance with all FERC, NERC, and SERC standards/requirements. 

Paul has more than 18 years of electric utility experience, primarily in transmission operations, planning, asset management and relay design.

Prior to Entergy, Paul worked for Waldemar S Nelson in New Orleans, LA where he provided engineering services to the oil and gas industry.

He has a bachelor's degree in electrical engineering from the University of New Orleans. He's a member of the Institute of Electrical and Electronics Engineers, Inc. and is a registered professional engineer in Louisiana.

When Wednesday, 10 April 2019, 16:30 - 17:30
Where Room 1202 Patrick F. Taylor Hall
More Hide Abstracts.   Announcement (PDF)
 
Title GPU Road Network Graph Contraction and SSSP Query
Speaker Roozbeh Karimi
ECE Doctoral Student
Abstract

A Single Source Shortest Path (SSSP) query on a weighted graph computes the distance (sum of weights) from a given start node to all other nodes in the graph. SSSP queries on large road-network graphs are used in a variety of applications. Existing GPU accelerated SSSP queries work well on most graph types, but on road-network graphs they have failed to deliver any advantage over the common CPU algorithms such as Dijkstra's. Acceleration of SSSP has been effective when a graph can be pre-processed on a CPU. PHAST is to date one of the fastest algorithms for performing SSSP queries on pre-processed road-network graphs. PHAST operates on graphs pre-processed in part using Geisberger's contraction hierarchy (CH) algorithm. This pre-processing is time consuming, limiting PHAST's usefulness when graphs are not available in advance. CH iteratively assigns scores to nodes, contracts (removes) the highest-scoring node, and adds shortcut edges to preserve distances. Iteration stops when only one node remains. Scoring and contraction rely on a witness path search (WPS) of nearby nodes. Little work has been reported on parallel and especially GPU CH algorithms. This is perhaps due to issues such as the validity of simultaneous potentially overlapping searches, score staleness, and parallel graph updates.

In this talk a brief overview CH and the challenges associated with a GPU implementation will be discussed and a GPU contraction algorithm, CUCH, is presented which overcomes these difficulties. CUCH exposes parallelism by partitioning a graph into levels composed of independent sets of nodes (non-adjacent nodes) with similar scores. This allows contracting multiple nodes simultaneously with little coordination between threads. A GPU-efficient WPS is presented in which a small neighborhood is kept in shared memory and a hash table is used to detect path overlap. Low-parallelism regions of contraction and query are avoided by halting contraction early and computing APSP on the remaining graph. A PHAST-like query computes SSSP using this contracted graph. Contraction of some DIMACS road network graphs on an Nvidia P100 GPU achieves a speedup of 20 to 37 over Geisberger's serial code on a Xeon E5-2640 v4. Query times on CUCH- and CH-contracted graphs were comparable.

Bio

Roozbeh Karimi is a Ph.D. candidate in the division of Electrical and Computer Engineering at Louisiana State University. He received his B.Sc. in Electrical Engineering from Azad University Tehran, and his M.Sc. in Biomedical engineering from Amirkabir University of Technology (Tehran Polytechnic), Iran. He worked as part of the LA-SiGMA research program between 2013 and 2015 working on optimizations on the Intel Xeon Phi platform. His research interests are in applications of massively parallel architectures specifically GPGPUs and development of algorithms for such platforms.

When Thursday, 25 April 2019, 12:30 - 13:30
Where Room 3285 Patrick F. Taylor Hall
More Hide Abstracts.   Announcement (PDF)
 
Title Robust Verification of Hybrid Systems
Speaker Pavithra Prabhakar
Kansas State University
Abstract

Cyber-physical systems (CPSs) consist of complex systems that combine control, computation and communication to achieve sophisticated functionalities as in autonomous driving in driverless cars and automated load balancing in smart grids. The safety criticality of these systems demands strong guarantees about their correct functioning. In this talk, we will present some of our work on formal verification techniques for cyber-physical systems analysis using the framework of hybrid systems. Hybrid systems capture an important feature of CPSs, namely, mixed discrete-continuous behaviors that arise due to the interaction of complex digital control software (discrete elements) with physical systems (continuous elements). We will focus on certain robustness properties of these systems, and present scalable techniques based on abstraction-refinement for their analyses.

Bio

Pavithra Prabhakar is an associate professor in the Department of Computer Science and Peggy and Gary Edwards Chair in Engineering at the Kansas State University. She obtained her doctorate in Computer Science and a masters in Applied Mathematics from the University of Illinois at Urbana-Champaign, followed by a CMI postdoctoral fellowship at the California Institute of Technology and a faculty position at the IMDEA Software Institute in Spain. Her main research interest is in formal analysis of cyber-physical systems with emphasis on both foundational and practical aspects related to automated and scalable techniques for verification and synthesis of hybrid systems. She is the recipient of a Marie Curie Career Integration Grant from the EU, a National Science Foundation CAREER Award and an Office of Naval Research Young Investigator Award.

When Tuesday, 26 November 2019, 10:00 - 11:00
Where Room 3316E Patrick F. Taylor Hall
More Hide Abstracts.   Announcement (PDF)
 
Title Mixed H2/H Control—A New Paradigm
Speaker Xiang Chen
University of Windsor, Windsor, Ontario
Abstract

A new design paradigm is discussed in this talk which allows control designs to achieve performance complements without trade-off. In particular, a revisit of the mixed H2/H control is presented with the proposed new control structure, which is motivated by the famous Youla-Kucera parameterization of all stabilizing controllers. It is shown that this new paradigm is not only able to automatically render the H2 control performance if there is no modeling mismatch for the plant, but also provide recovery, instead of compromise, of the optimal performance when the modeling error is present, noting that the compromise is normally seen in traditional mixed or multi-objective designs. It is also noted that the recovery of the robust performance is regulated by the “measured error size” of the modeling mismatch, hence, resulting in less conservativeness of the control performance. An inverted pendulum example is presented to validate the design expectations of the new paradigm.

Bio

Xiang Chen received M. Sc. and Ph.D. degree in system and control from Louisiana State University in 1996 and 1998. He held cross-appointed positions in Department of Electrical and Computer Engineering and Department of Mechanical, Automotive and Materials Engineering at the University of Windsor, Ontario, and is currently a Professor in the Department of Electrical and Computer Engineering. He has made fundamental contribution to Gaussain filtering and control, control of nonlinear systems with bifurcation, networked control system, and optimization of field sensing network. He has also made significant contribution to industrial applications of control and optimization in automotive systems and in visual sensing systems for manufacturing through extensive collaborative research and development activities with automotive, robotics, and manufacturing industries. Some of the deliverables have been patented by relevant companies or transferred to technological products of relevant companies. He is currently a Senior Editor for the IEEE/; ASME Transactions on Mechatronics, an Associate Editor for SIAM Journal on Control and Optimization, and Associate Editors for International Journal of Intelligent Robotics and Applications, Control Theory and Technology (English Version), and Unman Systems. He received the Award of Best Paper Finalist from 2017 IEEE/; ASME International Conference on Advanced Intelligent Mechatronics (AIM 2017), the Award of Best Student Paper Finalist (as the supervisor author) from 2015 ASME DSCC, the New Opportunity Awards from the Canadian Foundation of Innovation (CFI) and from the Ontario Centre of Excellence-- Materials and Manufacturing Ontario, as well as 4 times Research Awards from the University of Windsor. His research has been well supported by government agencies at both federal and provincial levels in Canada and from industrial companies in both Canada and USA. His current research interests include multi-objective complementary optimization and control of systems with complexities, optimization and control of field sensing network and field sensor based autonomous operations, graph-/game-theoretic approaches for complex networked systems, as well as control applications to automotive systems and autonomous vehicles. He is a registered Professional Engineer in Ontario.

When Thursday, 5 December 2019, 10:00 - 11:00
Where Room 3316E Patrick F. Taylor Hall
More Hide Abstracts.   Announcement (PDF)
 
Title Achieving Better Resource Scheduling through Enhanced OS and ML Technologies
Speaker Lei Liu
Chinese Academy of Sciences
Abstract

To improve resource utilization and meet the new challenges—microservices—in cloud environments, this talk will include the results from our Sys-Inventor lab in recent years. The contents include 1) the technical evolutions for memory partitioning approaches; 2) hybrid memory management in OS for systems using NVM-DRAM; and 3) Next Generation OS -- leveraging ML to build resource scheduler for microservices. Some of our results benefit both the computer industry and academia.

Bio

Lei Liu is an Associate Professor of Computer Science at Institute of Computing Technology (ICT), Chinese Academy of Sciences (CAS), where he leads the Sys-Inventor Lab, which is a part of the State Key Lab of Computer Architecture. He was a visiting scholar in CS at the University of Rochester (UR). He has led research projects that have advanced Memory Systems, OS, Performance Isolation, and Profiling. As the leading author, Dr. Liu has published research articles regarding the architecture and operating system in venues that include ISCA, PACT, IEEE TC, TPDS, ACM TACO, ICCD, and others. Recently, his work has focused on the intersection of the hybrid memory system (DRAM-NVM), OS for emerging technologies (including GPU, Graph, Quantum Computing as well as AI). Lei Liu has served as the PC/ERC members, Chairs for a number of mainstream conferences (e.g., HPCA, SC, PACT, ICS, IISWC, ISCA, ASPLOS, ICCD, ICPP, IPDPS, HPCC, General co-Chair for ACM ICS-2018) and reviewer for well-known Transactions (e.g., ACM TACO, ACM JETC, ACM TECS, ACM TRTS, IEEE TPDS).

When Tuesday, 10 December 2019, 10:00 - 11:00
Where Room 3285 Patrick F. Taylor Hall
More Hide Abstracts.   Announcement (PDF)