Energy Maps for Mobile Wireless Networks: Coherence Time versus Node Mobility

Date: 
Tuesday, January 9, 2007 - 11:11am


Prof. Volkan Rodoplu
Department of Electrical and Computer Engineering
University of California Santa Barbara
Date: Wednesday, January 17, 2007
Time: 1:00pm-2:00pm
Location: CTL (trailer 932 101)

Abstract:

The wireless networks of the next 10 years will consist of a
plethora of microprocessor-sensor units embedded in clothes, shoes,
cars, buses, as well as the more traditional portable handhelds, and
laptops. Today, information flows in wireless networks via a limited
number of wireless gateways. In the future, information is expected to
flow through thousands to millions of wireless devices themselves. Most
of these devices will be mobile and energy-limited, and will have to
make decisions on the fly on how to communicate information through
thousands to millions of other devices in between to reach destination
nodes, as well as gateways into the wired domain. It will no longer be
possible to track individual paths and individual nodes; hence, it is
essential that an aggregate view of the essential qualities of the
mobile network be built and be made available. Quality-of-Service (QoS)
decisions regarding energy consumption, delay, and throughput will still
play a prominent role in making intelligent decisions to conserve the
limited energy supply of devices, and meet delay and throughput
requirements in these future networks that consist of thousands to
millions of mobile, microprocessor-sensor devices.

With this vision, in this talk, we develop new methodologies for mobile,
large-scale wireless sensor networks. We propose a novel framework to
share, retain and refine end-to-end QoS metrics in the joint memory of
the nodes, over time scales over which this information can be spread to
the network and utilized for energy planning decisions. In analogy with
the point-to-point link concepts, we introduce the “coherence time” of
end-to-end QoS metrics for mobile wireless networks. We show that as
long as the coherence time of QoS metrics is much larger than the
“spreading period”, mobile wireless networks can track end-to-end QoS
metrics. This is a surprising conclusion given our current understanding
of mobile networks, which correlates tractability with the amount of
individual node mobility rather than the coherence time of QoS metrics.

As an example of this methodology, we construct “energy maps,” which are
maps of the end-to-end energy metrics in space. We show how to
(1) compute the spatial derivatives of energy potentials in mobile
networks,
(2) construct energy maps on-demand via path integration methods, and
(3) distribute, share, fuse, and refine energy maps over time by
information exchange during encounters. In order to put the energy maps
to use, we present an algorithm for energy optimization, based on the
energy maps, that finds the optimal bit allocation strategy to minimize
the energy consumption, subject to a delay constraint. We show that
significant energy savings are obtained by leveraging network mobility
and the energy maps, when compared with a competing algorithm that
allocates the traffic at a constant rate without utilizing the energy
map. These techniques show how future, large-scale, mobile wireless
sensor networks can be handled via new techniques, and how to generalize
physical layer concepts such as coherence time, to network-layer
concepts related to QoS issues.

Biography:

Volkan Rodoplu received his B.S. degree in Electrical
Engineering from Princeton University in 1996 and his M.S. degree in
Electrical Engineering from Stanford University in 1998. He worked for
Texas Instruments (Dallas, TX) in the summer of 1998, on multiuser
detection and interference cancellation algorithms, and for Tensilica,
Inc. (Santa Clara, CA) in 2000 and 2001, on turbo decoding algorithms
and architectures for reconfigurable processors. He received his Ph.D.
in Electrical Engineering from Stanford University in 2003 and
subsequently joined the Department of Electrical and Computer
Engineering at UCSB as an Assistant Professor.

His research areas span underwater networks, terrestrial wireless mobile
sensor networks, and applications of cooperative game theory to wireless
networks. His research investigates the theoretical limits of minimum
energy networks as well as the delivery of minimum energy networking
solutions.

He is the recipient of the NSF CAREER Award (2007), UC Regents’ Junior
Faculty Fellowship (2006), Department of Electrical Engineering
Outstanding Service Award at Stanford (2000), B.George B. Wood Legacy
Prize, and G. David Forney Prize (1996), and the John W. Tukey Award (1995).

Host: Wim van Dam