Adaptable Network Systems : Harnessing Nature's Architectures for Artificial Network Systems

Networks are fundamental structures in biology and technology. Despite their commonality, understanding how to design networks that combine structural robustness with adaptability remains a significant challenge. This doctoral dissertation addresses the aforementioned gap by investigating materials for adaptable network ensembles, posing the question: how can network architecture be designed to balance structural and functional robustness with reconfigurability, and how can biological strategies inspire such designs? To explore this, two complementary material systems were pursued, namely, a structurally static and a structurally reconfigurable network approach. In the static approach, t-ZnO networks were fabricated via direct ink writing, preserving 3D connectivity within a 2D polymer matrix. The resulting flexible, freestanding films function as sensors for ultraviolet light and sustain over 100 bending cycles, demonstrating that robust connectivity facilitates consistent device performance. Furthermore, these flexible sensors were coupled with memristive devices to realize memsensors. For the reconfigurable approach, metal/electrolyte/metal systems were developed. The electrolyte facilitated the growth of reconfigurable, conductive filaments between the metal electrodes. This system emulates synaptic plasticity. These dynamic filaments exhibit resistive switching in varying time and length scales. Furthermore, filament properties such as the ratio between the high and the low resistive states, and fractal dimension can be altered by the applied stimulus. The filaments exhibit properties such as self-organization governed by diffusion limited aggregation, distributed plasticity, robustness which gives rise to redundant pathways, and 3D connectivity. Additionally, this work examines the slime mold Physarum polycephalum, whose transport networks inspire efficient and adaptive computation. By juxtaposing biological strategies with varied material systems, this thesis concludes that robust static connections as well as reconfigurable connections together ensure stability and adaptability. This work demonstrates design principles for bio-inspired network architectures, for both neuromorphic hardware design and materials-based network design.

Supplementary resources

Rights

Use and reproduction:


CC BY-NC-ND 4.0

Please note that individual components of the publication may be subject to other licensing or copyright conditions.

Cite

Citation style:
Could not load citation form.