|
|
|
| Recent Research Highlights | ||
|
|
This page contains brief summaries of recent research conducted by CSI researchers. This content will change each time the page is loaded or you may view all summaries. Evaluation of an
Ultra-Wideband Propagation Channel Dr. Jean-Marc Cramer, Prof. Robert A. Scholtz & Prof. Moe Z. Win
Unlike narrowband signals, the received signal in an UWB system
often bears little resemblance to the signal driving the transmitter's
antenna. Waves reflecting off or penetrating through objects in
the channel can undergo significant filtering, and the antennas at both
the transmit and receive ends cause pulse-shaping that can vary with
direction of transmission and reception. The result is the
received pulse shape associated with a given path is dependent on that
path. This work provides a needed algorithm, called the Sensor-CLEAN
algorithm, for taking into account these special bandwith-dependent
effects, so that quantitative comparisons of the UWB channel can be made
with more narrowband results, and the performance of UWB communication
systems predicted. The algorithm was applied to measured indoor propagation data to
develop models for the time- and angle-of-arrival of UWB signals, which
combined with the Sensor-CLEAN method for processing measured data also
enables the future statitstical description of propagation environments
in other building architectures and geometries.
Wireless Resource Management
Prof. Mehmet Akar, Prof. Urbashi Mitra, Ph.D. Candidate Ayanendu Paul, & Prof. Michael G. Safonov
In this work, we use control theoretic methods including tools
of hybrid/switched systems, adaptive control, optimization, and dynamic programming to
develop distributed, fast and robust
resource allocation algorithms that take into account the effects
of channel fading and mobility, and achieve a tradeoff between the
satisfaction levels of the mobile users and the network operator,
thereby provide satisfactory service for the users while reducing
the burden on the network such as undesired switching between base
stations.
Extraction of Good Graphical Models of Codes
Prof. Keith M. Chugg & Ph.D. Candidate Thomas R. Halford
|
|
 USC Viterbi School of Engineering |
USC Electrical Engineering Department |
Ultra-wideband (UWB) radio uses
signals with fractional bandwidth (3dB bandwidth divided by center
frequency) greater than 25%. 
Dynamic resource allocation in wireless networks is important to
maintain reliable communication links between base stations and
mobile users. In order to achieve this objective, the transmitted
powers/rates, base station assignments, and allocated channels
may need to be updated as the mobile user moves in a cellular
environment or when a new call is admitted to the network.
Although these resources can be controlled individually, joint
allocation improves capacity and battery life, and decreases
interference. 
We distiguish between two modeling problems: model extraction and model construction. The construction problem has been largely addressed by the coding theory communtity: given a set of design parameters, there exist well-known methods for designing codes with graphical models that imply low-complexity, near-optimal decoding algorithms. The extraction problem, however, is largely unsolved. Given an arbitrary linear block code, there exist no known methods for extracting good graphical models. For specific codes, there do exist good graphical models. One such example is the tail-biting trellis-based model of the Golay code. Current known good graphical models of specific codes depend on particular algebraic properties of those codes and do not extend to arbitrary codes. We are developing heuristics for the systematic extraction of good graphical models of arbitrary linear block codes and also in studying the fundamental tradeoff between the algebraic structure of a code and the complexity of allowable graphical models with a given topology.