AN INTRODUCTION TO NANOCOMPUTING - Bio-Inspired and Nanoscale Integrated Computing - page 16

Figure 1.11. Example of a simple cellular automaton. Each cell can be ''on'' (black)
or ''off'' (white). Depending on its neighbors, a cell may turn on or off in the next
time-step. This specific example shows a ''glider,'' a structure that emerges from
rules defined by the famous ''Game of Life'' by John Conway [25]. As time goes
forward, the glider moves diagonally. More complex cellular automata can
compute anything that a traditional computer can, and possibly more.
a computer becomes much easier because the same nanoscale structure can be
repeated in a regular pattern. Most of the ideas in this paradigm are also fault-
tolerant, that is, able to handle a few defective cells or neurons. This alleviates the
problem of variability when fabricating nanoscale devices. Their regular structure
usually implies that they can perform different functions depending on the
context, instead of being permanently hard coded with a fixed function; in other
words, such computers are highly reconfigurable. Most importantly the potential
computing power that is available with emergent properties is immense and only
beginning to be explored.
Even with the limited understanding of emergent properties, this paradigm
already has many applications. Cellular automata and cellular nonlinear networks
have been mostly used for image processing applications due to the highly parallel
nature and intuitive correspondence between each simple processor and pixels on
an image. Neural networks have been extensively studied for their applications in
artificial intelligence and are useful practically anywhere uncertainty is encoun-
tered in computation. It has also been shown theoretically that cellular automata
and neural networks can do anything that today's computers can do [26, 27]. With
all this in mind, one of the main challenges of this paradigm is to find a way to
harness emergent properties for a wider variety of applications.
Nanoscale devices that implement cellular automata, cellular nonlinear net-
works, and neural networks are actively being researched. NanoCells, quantum-
dot cellular automata (QCA), and RTDs, described in the previous section, are
just a few possible ways to realize this paradigm. Quantom-dot cellular automata
are discussed in Chapter 4, and nanoscale neural networks are discussed in
Chapter 17.
1.5.2. Wave Computing
Waves are an elegant but complicated way to communicate and manipulate
abstract data. One of the most powerful features of waves is the phenomenon of
diffraction, or the behavior of light as it propagates around objects or through a
nonuniform medium. Perhaps the best known example of the power of diffraction