Andrea Goldsmith - US grants

This is a collaborative research project between UC Berkeley and Stanford University to develop a validated analytical methodology for designing control systems where some of the feedback loops are closed over wireless communication links. The proposed work is an interdisciplinary effort between two very disparate research areas: wireless communications and control theory. This work will have a great impact on many fields where wireless communication is beginning to see an increasing presence such as transportation, manufacturing and remote sensing applications.

Most current controller designs assume that communication between the sensors and actuators and the central logic system is done over hard-wired lines providing essentially perfect transmission. Thus, the controller design typically assumes no rate constraints, delay, or loss of information. However, there are many applications where this communication must take place over imperfect wireless links where these imperfections can lead to loss of performance and even instability. There currently exists no unified theory for the design of closed loop control systems using wireless links. This proposal is aimed at developing such a theory.

The first stage of the proposed work will focus on characterizing the communications network carrying the sensor, actuator, and control information in terms of its rate constraints, and delay and packet loss statistics. The next stage will investigate joint optimization of the controller and communication system.

The research results will be presented at both communications and controls conferences. The final results will be incorporated into advanced graduate courses. A very important aspect of this project is the involvement of undergraduate students in the research. Both UC Berkeley and Stanford have active UROP (Undergraduate Research Opportunities Programs). The area of communication/control is very attractive to undergraduates and it is anticipated that undergraduates will participate in the project due to its interdisciplinary nature and its potential to impact real world problems.

Center Name: Center for Science of Information
Center Director: W. Szpankowski,
Lead Institution: Purdue University

The foundations of modern communications and ancillary trillion plus dollar economic windfall were laid in 1948 by Claude Shannon who introduced a general mathematical theory of the inherent information content in data and its reliable communication in the presence of noise. While Shannon?s Theory has had a profound impact, its application beyond storage and point-to-point communication, e.g., to the Internet, poses fundamental challenges, among the most vexing facing today?s scientists and engineers. The overarching vision of the proposed Center for Science of Information is to develop a new science of information that incorporates common features generally associated with data/information, such as space, time, structure, semantics and context that are not addressed by Shannon?s Theory. The realization of this vision requires a center-level environment that can focus the efforts of a sizeable (and diverse) group of researchers, for a protracted period of time, on these critically important challenges, which could have far reaching societal impact and enormous economic ramifications. Under the umbrella of this overarching vision, the proposed center will explore the following fundamental issues: (i) modeling complex systems and development of analytical techniques for information flow (e.g., understanding Darwinian selection); (ii) quantification and extraction of informative substructures in complex systems (e.g., discovering functionally relevant structures in gene regulatory networks or modular entities in social networks); (iii) understanding of spatio-temporal coding used to exchange information through timing and localization in complex systems (e.g., building more efficient ad hoc networks and understanding neuronal activity); (iv) data-driven knowledge discovery based on formal information-theoretic measures (e.g., finding semantically relevant information in unstructured repositories); (v) steganography, data obfuscation and hiding as mechanisms for robustness (e.g., developing secure systems for monitoring and surveillance); and (vi) discovering principles of redundancy and fault tolerance in diverse natural systems (e.g., understanding the interplay between erasure coding and distributed system design).

The intellectual merits of the proposed center include the community of students and academic and industrial scholars it seeks to sustain, the theoretical advances it hopes to achieve, and the novel insights and tools it hopes to provide to explicate a myriad of diverse systems, ranging from the life sciences through business applications. The broader impacts of this Center extend beyond the potential scientific, societal and economic ramifications and include the creation of an ?active and thriving community of students and scholars? who will train the next generation of scientists and engineers, enlighten the public, and ultimately pave the way for the next information revolution. The Center team is composed of over 40 investigators, many having already made significant accomplishments in multiple research areas relevant to the Science of Information. The Center team is a very diverse group: it has a mix of junior and senior researchers, including several members of underrepresented groups. They bring expertise in all essential areas of research, including Computer Science, Chemistry, Economics, Statistics, Environmental Science, Information Theory, Life Sciences, and Physics. The institutional partners include nine premier institutions (Purdue, Bryn Mawr, Howard, MIT, Princeton, Stanford, UC Berkeley, UCSD, and UIUC), two of which have significant underrepresented student populations. The academic institutions are complemented by the Center?s industrial partners (Amgen, Bell Labs, Configuersoft, Google, HP, Lilly, NEC, Qualcomm, and Yahoo) and by world-renowned researchers at international institutions.

This project focuses on the problem of information acquisition, state estimation and control in the context of cyber physical systems. In our underlying model, a (set of) decision maker(s), by controlling a sequence of actions with uncertain outcomes, dynamically refines the belief about stochastically time-varying parameters of interest. These parameters are then used to control the physical system efficiently and robustly. Here the cyber system collects, processes, and acquires information about the underlying physical system of interest, which is used for its control. The proposed work will develop a new theoretical framework for stochastic learning, decision-making, and control in stochastically-varying cyber physical systems.

In order to obtain analytical insights into the structure of efficient design, we first consider the case where the actions of the cyber system only affect the estimate of the underlying physical system. This class of problems arises in the context of (distributed) sensing/tracking of a physical system in isolation from cyber system control of the physical system's state. Joint state estimation and control for cyber-physical systems will then be considered. Here the most natural first step is to obtain sufficient conditions and/or special classes of systems where a separated approach to the information acquisition and efficient control is (near) optimal. To demonstrate its utility in practice, our theoretical framework will be applied in the specific context of energy efficient control of data centers and robust control of the smart grid under limited sensing.

The intellectual merit of this work will be to develop a theoretical framework for the design of cyber-physical systems including information acquisition, state estimation, and control. In addition, separation theorems for the optimality of separate state estimation and control will be explored.

In terms of broader impacts, significant performance improvement of control systems closed over communication networks will impact a wide range of applications for societal benefit, including smart buildings, intelligent transportation systems, energy-efficient data centers, and the future smart-grid. The PIs plan to disseminate the research results widely through conferences and journals, as well as by organizing specialized workshops and conference sessions related to cyber physical systems. The proposed project will train Ph.D. students as well as enrich the curriculum taught by the PIs in communications, stochastic control, and networks. The PIs have a strong track record in diversity and outreach activities, which for this project will include exposure and involvement of high school and undergraduate students, including under-represented minorities and women.

Many of the sophisticated electronic devices ubiquitous today have been enabled by digital signal processing (DSP) that converts the analog signals associated with the physical world to the digital domain. Moore's law allows such digitized signals to be processed and communicated with small, low-cost energy-efficient digital hardware. Yet fundamental questions regarding the performance limits and tradeoffs associated with the sampling, quantization, communication, and reconstruction of analog data with respect to both fidelity and overall energy consumption remain open. Better understanding of these performance limits and tradeoffs will significant enhance capabilities for collection, processing, and communication of analog sensor data beyond the current state of the art. These capabilities are particularly acute for emerging sensor network applications in health and wellness, security, energy-efficient infrastructures, and smart cities, where many low-cost low-energy analog sensors will be collecting large amounts of data and transmitting it to remote locations for processing.

The proposed research will investigate the interplay between sampling, quantization, communication and reconstruction of analog signals under memory constraints, communication constraints, and energy constraints. The goal of the proposed work is to develop a fundamental rate-distortion theory for the sampling, quantization, and reconstruction of analog data subject to these constraints. Specific Shannon-theoretical limits for the performance of combined sampling and source coding - the tradeoff between the digitizer's sampling rate and quantization precision in terms of distortion will be investigated. In addition, the optimal fidelity in the reconstruction of analog data from a sequence of samples will be derived. Fundamental limits on communication under energy constraints, where an analog sensor must digitize and communication its data under finite energy constraints, will be determined. Methods used will draw from prior work in Shannon capacity, rate-distortion theory, joint source-channel coding and separation, sampling, estimation, and statistics. In particular, after extending existing results on rate-distortion theory to incorporate sub-Nyquist sampling, joint source-channel coding techniques applied to the sampled analog data transmitted over a rate-limited channel will be developed. Energy constraints will be incorporated based on information-theoretic models for computation energy along with recent results on finite-block-length codes and minimum energy-per-bit capacity.