;news
Science;|;DOI: 10.1145/1953122.1953128
Kirk;L.;Kroeker
Biology-inspired
networking
Researchers have developed a new networking algorithm,
modeled after the neurological development of the fruit fly,
to help distributed networks self-organize more efficiently.
The frUit fly’S neurological development is serving as a model for a networking algorithm that research- ers say has the potential to
obsolesce traditional methods for
determining peer relationships in
distributed networks. Without prior
knowledge of how cells are connected, the fly’s developing nervous system allocates certain cells as leaders
that provide direct connections with
other nerve cells. That development
process, the researchers say, is similar
to conventional schemes used to manage distributed networks but is much
simpler and more robust than anything humans have yet devised.
PhotograPh Courtesy oF CarnegIe Mellon unIversIty
The lead scientist of the project to
develop the new networking algorithm
is Ziv Bar-Joseph, an associate professor of machine learning and computational biology at Carnegie Mellon
University. Bar-Joseph says he became
interested in biology-inspired computing while studying at Massachusetts
Institute of Technology during the time
of the initial sequencing of the human
genome. “I was fascinated by the opportunities this information presented
for improving human health and our
in this microscope image of the pupal stage of fruit fly development, nerve cells that become
sensory organ precursors (soPs) are identified by arrows. These cells send chemical signals
to neighboring cells, blocking them from becoming soPs.
understanding of the inner workings of
cells in our body,” he says. “However, I
also realized that the explosion of data
coming out of biology would require
the development of new computational
methods to make sense of this data.”
More than a decade later, the need
for sophisticated methods for dealing
with increasingly large sets of biologi-
cal data and even new types of infor-
mation remains a salient characteris-
tic of both computational biology and
