|
Topics
CAS Synthesis
The BISON project will study a small number of biology-inspired
Complex Adaptive Systems (CAS), applied to the technological niche
of dynamic networks, with the aim of elucidating principles or
regularities in their behavior. BISON seeks to develop a
rigorous understanding of why a given CAS does or does not
perform well for a given technological problem. A systematic
study of the rules governing good performance of CAS offers a
bottom-up opportunity to build more general understanding of the
rules for CAS behavior. The ultimate goal of the BISON project
is then the ability to synthesize a CAS that will perform well
in solving a given technological task based on the accumulated
understanding of its regularities when applied to different
tasks. The achievement of this goal would enable the systematic
exploitation of the potential of CAS, freeing technologists from
having to comb through nature to find the desired behavior. The
mechanisms of nature will continue to be a rich resource of
ideas and inspiration.
In addition to this ambitious overall objective, BISON has more
concrete objectives to obtain robust, self-organizing and
self-repairing solutions to important problems that arise in
dynamic networks at both the system layer and the application
layer. Here we outline these objectives.
CAS techniques for basic network services
Despite the current domination of fixed infrastructures, future
networks are expected to be increasingly dynamic. Heterogeneous
populations of mobile, wireless devices will form Ad-Hoc
Networks (AHN) with the goal of cooperation on specific tasks,
exchanging information or simply interacting informally to relay
information between themselves and the fixed network. AHN imply
not only mobility and wireless connections, but also frequent
joining and leaving of nodes, often changing interconnection
patterns and the possibility of multihop routing among mobile
nodes. Multihop routing offers numerous benefits: it extends
the range of a base station; it allows power saving; and it
allows wireless communication, without the use of base stations,
between users users located within a limited distance of one
another. Routing in AHN presents challenging technical
problems. One needs routing strategies that are flexible,
adaptive, and decentralized - classical techniques, based on
optimization with global knowledge, are not suitable for a
dynamic AHN. Instead, AHN routing must be based on local
knowledge, using decentralized control mechanisms, and be
capable of adapting rapidly to changing network conditions. In
contrast to routing which finds paths, the search function seeks
resources located at the nodes of a dynamic network. Discovery
is a form of search where there is partial information on what
is being sought such that the outcome results in what is
available. BISON will develop novel algorithms for routing,
searching, discovery, and monitoring in AHN, based on techniques
inspired by social insects and immune systems.
CAS techniques for advanced network services
Whereas Ad-Hoc and Virtual Networks are instances of dynamic
networks at the system layer, P2P and Grid systems are based on
dynamic networks at the application layer. Here, the canonical
problem to be solved is resource sharing, which may be
computation, storage or content. Developing P2P and Grid
applications is a complex task. First of all, they must be based
on completely decentralized control. Furthermore, these
applications reach out to harness the outer edges of the network
and consequently involve scales that were previously
unimaginable. Finally, the environments in which they are
deployed exhibit extreme dynamism in topology and load. BISON
will exploit ideas and techniques derived from CAS to support
the construction of robust, self-organizing and self-repairing
P2P and Grid applications capable of dealing with these issues.
In particular, we will focus our efforts in the development of
CAS approaches for P2P document sharing and load-balancing
schemes for Grid computing.
|