Self-Assembled Networks: Control vs. Complexity

DNA-based self-assembly of nanoelectronic devices is an emerging technology that has the potential to enable tera-to peta-scale device integration. However, self-assembly currently is limited to manufacturing small computing blocks (nodes) which must then be interconnected to build a larger computing system. In this paper, the authors study node networks created by varying control over three aspects of the self-assembly process (node placement, node orientation, and inter-node link creation). In particular, the authors examine the tradeoff between node complexity and control required during self-assembly to maximize the number of connected nodes in the network. As the level of control decreases, the authors find that node communication hardware needs to be augmented to allow link sharing between several transceivers. This also results in better network connectivity in the presence of defective nodes and links. Finally, the authors show that for a data parallel architecture with enough available nodes, the specific network topology has a negligible effect on performance