In the modernization of infrastructure systems spanning energy, transportation, and health we are seeing the convergence of big data analytics, cyber-physical systems, and the internet of things. The resulting societal-scale cyber-physical systems (S-CPS) provide new opportunities for efficiency yet expose novel vulnerabilities. In the energy systems, for example, the availability of streaming data from smart metering enables monetization of energy savings. These savings can be realized by employing novel variants of machine learning algorithms to generate energy analytics that allow customization of offerings to consumers and by creating the economic incentives necessary for investment in the instrumentation of physical infrastructure. On the other hand, these emerging service models depend on the underlying CPS infrastructure and thus, reveal new vulnerabilities due to ubiquitous sensing, real-time constraints, and ‘closing-the-loop’ attributes. This efficiency-vulnerability tradeoff is a fundamental challenge facing S-CPS, wherein scarce resources must be allocated amongst competitive agents with misaligned goals. To manage this tradeoff, a coordinator can provide incentives to align these goals, for instance, by ensuring the equilibrium behavior optimizes a societal cost.
In this talk, I present an algorithm for synthesizing incentive strategies that lead to efficient behavior when the preferences of the underlying agents are unknown to the coordinator and must be learned. In support of the incentive design and learning steps in the algorithm, I present an intrinsic characterization of Nash equilibria that is amenable to computation. This learning and mechanism design procedure aims to bridge the gap between the non-cooperative Nash equilibrium and a more efficient, perhaps socially optimal, solution. A coordinator can leverage the underlying CPS infrastructure to enhance system efficiency at the expense of revealing vulnerabilities. By focusing on the demand-side of the power grid, I provide tools for analysis of consumer privacy and the design of economic mechanisms for balancing the efficiency-vulnerability tradeoff. The combination of data-driven models and game-theoretic tools I present provides the foundation for a systems theory of S-CPS.
The next generation of complex engineered systems will see an unprecedented integration of electromechanical components, communication, and embedded computation. Imminent examples include self-driving vehicles, smart buildings, and UAVs for automated delivery of goods. It is critical that these new technologies be safe and efficient, as their failure would be socially and economically catastrophic.
This talk will focus on the challenge of integrating data-driven optimization algorithms into safety-critical control systems. The problem of selecting a suitable algorithm for use in large-scale optimization is currently more of an art than a science; a great deal of expertise is required to know which algorithms to apply and how to properly tune them. Moreover, there are seldom performance or robustness guarantees.
Our key observation is that iterative optimization algorithms may be viewed as discrete-time controllers, and the problem of algorithm selection/tuning may be viewed as a robust control problem. This viewpoint allows us to treat both electromechanical and algorithmic components in a unified manner. By solving simple semidefinite programs, we can derive robust bounds on convergence rates for popular algorithms such as the gradient method, proximal methods, fast/accelerated methods, and operator-splitting methods such as ADMM. Finally, our framework can be used to search for algorithms that meet desired performance guarantees, thus establishing a new and principled methodology for algorithm design. As an illustrative example, we synthesize a new family of first-order algorithms that explore the trade-off between performance and robustness to noise.
Imagine you are a brain-in-a-vat that wakes up connected to an unknown (robotic) body. You are connected to two streams of uninterpreted observations and commands. You have zero prior knowledge on the body morphology, its sensors, its actuators, and the external world. Would you be able to “bootstrap” a model of your body from scratch, in an unsupervised manner, and use it to perform useful tasks? This bootstrapping problem sits at the intersection of numerous scientific questions and engineering problems.Biology gives us a proof of existence of a solution, given that the neocortex demonstrates similar abilities.
I am interested in understanding whether the bootstrapping problem can be formalized to the point where it can be solved with the rigor of control theory. I will discuss a tractable subset of the set of all robots called the “Vehicles Universe”, which I consider a updated version, with modern sensors, of Braitenberg’s Vehicles. I will show that the dynamics of three “canonical” robotic sensors (camera, range-finder, field sampler) are very similar at the “sensel” level. I will present classes of models that can capture the dynamics of those sensors simultaneously and allow exactly the same agent to perform equivalent spatial tasks when embodied in different robots.
Schuller to use the award to study how light interacts with certain materials, particularly those with complex and asymmetric molecular arrangements, such as plastics.
“Getting the CAREER Award is a great honor,” said Schuller. “It’s a great validation for me and my work as a young researcher.”
The award, which amounts to $500,000 over five years, will allow Schuller and his research group to examine the interactions between light and possible alternative semiconducting materials. Whereas conventional photonic (light-manipulating) materials such as silicon crystals tend to exhibit uniform optical behaviors in all directions (isotropic), other materials, including plastics, have optical properties that differ by direction (anisotropic).
Schuller’s research group will focus on examining the complex optical properties of organic (carbon-based) materials such as plastics. Their findings could in turn lead to developments that could enhance the performance of organic photonic devices. Additionally, the research could open new doors to the manufacture of low-cost, lightweight and flexible semiconductors that can harness and manipulate light for various applications.
The goal in networked control of multiagent systems is to derive desirable collective behavior through the design of local control algorithms. The information available to the individual agents, either through sensing or communication, invariably defines the space of admissible control laws. Hence, informational restrictions impose constraints on achievable performance guarantees. The first part of this talk will provide one such constraint with regards to the efficiency of the resulting stable solutions for a class of networked resource allocation problems with submodular objective functions. When the agents have full information regarding the resources, the efficiency of the resulting stable solutions is guaranteed to be within 50% of optimal. However, when the agents have only localized information about the resources, which is a common feature of many well-studied control designs, the efficiency of the resulting stable solutions can be 1/n of optimal, where n is the number of agents. Consequently, such schemes in general cannot guarantee that a system comprised of n agents can perform better than a system comprised of just a single agent. The second part of this talk will focus on identifying how augmenting the information to the agents can impact achievable performance guarantees. While providing the agents with additional information can lead to control designs with improved efficiency guarantees, it turns out that such gains frequently come at the expense of the underlying convergence rates. Hence, there is an apparent tradeoff between short-term and long-term performance guarantees in multiagent systems and we will characterize this tradeoff in a simple distributed graph coloring problem. The last part of this talk will present some preliminary results on robust mechanisms for social coordination.
Endowment named after Herbert Kroemer, UCSB emeritus professor of Electrical and Computer Engineering and of Materials and 2000 Nobel Laureate
The development of the bright white light-emitting diode (LED) signaled the beginning of the end for the incandescent bulb, which, at only five percent efficiency, emits far more in heat than light.
But, even with the LED’s phenomenal 50 percent efficiency, can the cooler-burning, longer-lasting LED bulb be made even better? UC Santa Barbara materials professor Chris Van de Walle thinks it might be possible. And that is the kind of research he looks forward to pursuing as the first person named to UCSB’s newly established Kroemer Chair in Materials Science.
“I’m extremely honored,” said Van de Walle. “I’m a great admirer of Herbert Kroemer, and I feel very privileged to be chosen to be the inaugural recipient of the chair that bears his name.”
To decarbonize the electric power grid, there have been increased efforts to utilize clean renewable energy sources, as well as demand-side resources such as electric loads. This utilization is challenging because of uncertain renewable generation and inelastic demand. Furthermore, the interdependencies between system states of power networks or interconnected loads complicate several decision-making problems. In this talk, I will present two control and optimization tools to help to overcome these challenges and improve the sustainability of electric power systems. The first tool is a new dynamic contract approach for direct or indirect load control that can manage the financial risks of utilities and customers, where the risks are generated by uncertain renewable generation. The key feature of the proposed contract method is its risk-limiting capability, which is achieved by formulating the contract design problem as mean-variance constrained risk-sensitive control. I will present a dynamical system approach to track and limit risks. The performance of the proposed contract framework is demonstrated using data from the Electricity Reliability Council of Texas. The second tool is developed for combinatorial decision-making under system interdependencies, which are inherent in interconnected loads and power networks. For such decision-making problems, which can be formulated as optimization of combinatorial dynamical systems, I will present a linear approximation method that is scalable and has a provable suboptimality bound. The performance of the approximation algorithm is illustrated in ON/OFF control of interconnected supermarket refrigeration systems and power network topology optimization. Finally, I will discuss several future research directions in the operation of sustainable systems, including a unified risk management framework for electricity markets, a selective monitoring and control mechanism for resilient power grids, and contract-based modular management of cyber-physical systems.
Control of transportation networks remains a challenging problem despite recent advances in the engineering of cyber-physical systems. A common feature of these systems is the propagation of nonlinear dynamics over interconnected network components. As a result, these systems exhibit complex global behavior such as large-scale congestion in traffic flow networks. The ongoing advancements in automated vehicles and infrastructure operations will further influence traffic flow dynamics and alter the global network behavior. Motivated by these challenges, this talk will focus on a class of analysis and control synthesis techniques for transportation networks.
First, intrinsic properties of traffic flow dynamics will be exploited to derive a new structural property for transportation networks. This “mixed monotonicity” property is viewed as an extension of the classical notion of monotonicity in dynamical systems. Second, it will be shown that mixed monotonicity enables efficient finite state abstraction of traffic flow dynamics, which allows for correct-by-construction synthesis of control strategies. Third, an approach to analyze the dynamical behavior of large-scale transportation networks will be presented. This approach relies on the embedding of mixed monotone dynamics into a higher dimensional system. Finally, future directions for the engineering of cyber-physical systems in transportation networks will be discussed.
This talk presents a systematic study of synchronization on distributed (networked) systems that spans from theoretical modeling and stability analysis to distributed controller design, implementation and verification. We first focus on developing a theoretical foundation for synchronization of networked oscillators. We study how the interaction type (coupling) and network configuration (topology) affect the behavior of a population of heterogeneous coupled oscillators. Unlike existing literature that restricts to specific scenarios, we show that phase consensus (common phase value) can be achieved for arbitrary network topologies under very general conditions on the oscillators’ model.
We then focus on more practical aspects of synchronization on smart grids. We propose a load-side frequency control scheme that can rebalance power and resynchronize frequencies after a disturbance (primary control), while restoring the frequency to its nominal value (secondary control). Unlike the generation-side secondary frequency control that is centralized, our load-side control only requires each bus to communicate with its neighbors. Furthermore, our scheme also provides a fair distribution of the load corrections by minimizing the total disutility of the controllable loads, and is able to preserve inter-area flows and thermal limits. We prove this distributed load-side control is globally asymptotically stable and further illustrate its convergence with numerical simulations.
21st International Symposium on High-Performance Computer Architecture (HPCA) is one of the top computer architecture conference
HPCA provides a high-quality forum for scientists and engineers to present their latest research findings in this rapidly-changing field. This year 51 papers were accepted out of 228 submissions. Xie and Li’s paper titled “Architecture Exploration for Ambient Energy Harvesting Nonvolatile Processors” was selected as the Best Paper Award Winner. This work is a collaboration with Penn State University and Tsinghua University.
Shuangchen Li is a first-year Ph.D. student in ECE department at UCSB. He received his MS and BS degree from Tsinghua University.