|
Abstract
The critical importance of providing communication service in the face of
failures has been recognized in the public switched telephone network, and a
great deal of attention has been paid to making these networks survivable
and self-healing. Although the past decade has seen an increase in the
deployment of wireless access networks and an exponential growth rate in the
number of users, little emphasis has been placed on understanding or
improving the survivability of these networks. As the public's demand for
and dependence on mobile services increases, users will ultimately demand
the same reliable service guarantees that are characteristic of wireline
telecommunications and data networks.
Our overall objective is to develop a comprehensive treatment of
survivability for wireless access networks. This talk will discuss our
initial focus on radio-level survivability of cellular and PCS networks
which is two-fold. Our first focus is on network design for survivability.
We propose cellsite architectures that provide dual-homing to mobile users
and multilayer adaptive resource management protocols that dynamically
redistribute radio resources on an as-needed basis. The examples described
include adaptive channel allocation and admission control algorithms.
Our second focus is on survivability analysis. We are working to define
real-time metrics to characterize and identify the critical states of mobile
network performance in the wake of channel failures, congestion, signal
degradation, etc. We use these metrics to define a survivability index as a
cost/performance function. We use our discrete-event simulation tools for
comparative analysis of proposed algorithms. As these algorithms need to
operate at multiple layers of the communications architecture, proper
evaluation of such techniques must take into account a variety of simulation
scenarios, parameterized by a large number of variables that need to be
monitored in real-time. To handle this state-space explosion, we are using
information visualization techniques to understand the complex spatial and
temporal relationships between performance and cost metrics that influence
adaptive algorithms. The dynamic visualizations of system variables and
interactions among them are fundamental to leveraging the user's intuition
and domain expertise to help explore (and reduce) a large search space of
these variables.
Biography:
Teresa A. Dahlberg received a B.S. in Electrical Engineering from the
University of Pittsburgh in 1984 and an M.S. and Ph.D. in Computer
Engineering from the North Carolina State University in 1990 and 1993.
Teresa worked as a hardware and software developer with the IBM Corp. for
ten years prior to joining the Electrical and Computer Engineering
department at the University of North Carolina at Charlotte as an assistant
professor in 1995. When the university created the College of Information
Technology in July 2000, Teresa moved to the Computer Science department in
the new college. Her research interests include wireless and mobile
networks, adaptive algorithms, fault-tolerance, as well as simulation and
information visualization methods of analysis.
| 
