Select a calendar:
Filter February Events by Event Type:
SUNMONTUEWEDTHUFRISAT
Events for February 28, 2012
-
Theoretical Foundations of Multi-Core Systems Design: A Dynamical Systems Perspective
Tue, Feb 28, 2012 @ 10:30 AM - 11:30 AM
Ming Hsieh Department of Electrical and Computer Engineering
Conferences, Lectures, & Seminars
Speaker: Paul Bogdan, Carnegie Mellon University
Talk Title: Theoretical Foundations of Multi-Core Systems Design: A Dynamical Systems Perspective
Abstract: Recent advances in CMOS technology enable the integration of tens and soon thousands of heterogeneous processing cores communicating via the Networks-on-Chip (NoC) paradigm. Many of the modeling and optimization approaches based on queuing theory have ignored important traffic characteristics (e.g., non-stationarity, fractality) that can pose serious challenges such as buffer overflows or deadline missing. Under these circumstances, it is crucial to take into account traffic characterization for both dynamic and static NoC optimization.
In this presentation, I will show that a statistical physics inspired approach designed to capture NoC traffic characteristics via a dynamical master equation can provide a viable solution for solving the above-mentioned problems. Compared to existing Markovian models, this radically new approach facilitates a more accurate evaluation of various performance metrics and reshapes the space models used for the online optimization of communication infrastructures. Starting from fractal state-space models of NoC traffic, I formulate the power and peak temperature management of heterogeneous NoCs as a constrained finite horizon fractal optimal control problem. Towards this end, I show that fractal characteristics can be accounted for via fractional state space models and that the online controller can be efficiently synthesized via linear programming. This approach not only contributes to significant power savings, but it also opens new avenues for the dynamic optimization of large-scale systems exhibiting fractal dynamics.
This statistical physics approach to dynamic processes taking place on networked architectures has many practical applications ranging from multi-core systems power management, to bacteria propelled micro-robotic swarms, green transportation via vehicular traffic optimization, and regenerative medicine.
Biography: Paul Bogdan received his BSc degree in Automatic Control and Computer Science from the âPolitehnicaâ University of Bucharest and his Ph.D. degree in Electrical and Computer Engineering from Carnegie Mellon University, Pittsburgh. He is a Post-Doctoral Fellow in the Electrical and Computer Engineering Department at Carnegie Mellon University. He was awarded the Roberto Rocca PhD Fellowship. His research interests include performance analysis and design methodologies for multicore systems, the theoretical foundations of cyber-physical systems, the modeling and analysis of bio-inspired computing, and the applications of statistical physics to biological systems and regenerative medicine.
Host: Massoud Pedram
Location: Hughes Aircraft Electrical Engineering Center (EEB) - 248
Audiences: Everyone Is Invited
Contact: Annie Yu
-
Low-Complexity Equalization for Multi-scale Multi-lag OFDM Channels
Tue, Feb 28, 2012 @ 03:00 PM - 04:00 PM
Ming Hsieh Department of Electrical and Computer Engineering
Conferences, Lectures, & Seminars
Speaker: Zijian Tang, Research Scientist of TNO the Hague, the Netherlands
Talk Title: Low-Complexity Equalization for Multi-scale Multi-lag OFDM Channels
Abstract: We consider an orthogonal frequency-division multiplexing (OFDM) transmission scheme over wideband underwater acoustic channels, where the propagation paths can experience distinct Doppler effects (manifested in signal scales) and time of arrivals (manifested in delays). We capture such an effect in this paper with a multi-scale multi-lag (MSML) model, and show that the resulting frequency-domain MSML-OFDM channel is subject to inter-carrier interference (ICI), whose amount differs per subcarrier. The corresponding channel matrix can still be approximated as highly sparse, but lacks a specific structure that can optimally be exploited by those low-complexity equalizers proposed for narrowband channels. In this paper, we propose to use the conjugate gradient (CG) algorithm to equalize the channel iteratively. The suitability of the preconditioning technique, that often accompanies the CG to accelerate the convergence, is discussed for the MSML-OFDM channel. We show that in order for the preconditioner to function properly, optimal resampling is indispensable.
Biography: Zijian Tang received the MSc in electrical engineering and the PhD degree from the Delft University of Technology (TU Delft) in 2003 and 2007, respectively. After working with Mathworks for one year, he joined the Sonar Group at Netherlands Organisation for Applied Scientific Research (TNO) in 2008 as a research scientist, where the research area covers underwater acoustic communications, anti-submarine warfare, marine mammal detection etc. Since 2010, he has also been holding a (part-time) research fellow position at the Network & Circuits Group of TU Delft. Zijian Tang is the recipient of the best student paper award (honourable mention) at ICASSP 2007, and was nominated to the Simon Stevin Gezel prize in 2008.
Host: Prof. Urbashi Mitra, ubli@usc.edu, x04667
Location: Hughes Aircraft Electrical Engineering Center (EEB) - 539
Audiences: Everyone Is Invited
Contact: Gerrielyn Ramos
