Temporal Network Characteristics for Bio-Inspired Networks

Bio-inspired algorithms offer a heuristic approach to optimizing complex systems, and have become increasingly important within a variety of technical domains. In prior work, we developed a hybrid Genetic-Algorithm-Voronoi approach to create adaptive networks that optimize coverage within noisy, obstacle-rich finite spaces [1]. Performance of these networks was then categorized in terms of application-driven measures, like Percent Area Coverage and Cumulative Distance Traveled. In more recent work [2], we approached the analysis of these swarm-like networks using a Temporal Network Graph (TNG) framework, where networks are allowed to be changing over time to model dynamical changes in the network structure. This allowed us to utilize a variety of network-centric measures in a time-dependent matter, giving new insight into how these adaptive networks evolve in a variety of environmental conditions. Specifically, we used an edge-centric representation of the network, and tracked how eigenvector centrality and regularity changed over the course of network deployment. It is shown that the distribution of edge lengths undergoes a phase transition like behavior, and our cross-correlation analysis of time traces shows similar results.

Supported in part by UAE ICTFund and Colgate University Undergraduate Research Fund.