Hardware Implementation of a Biomimicking Hybrid CA

Abstract

A hybrid model, combining a Cellular Automaton (CA) and a multi-agent system, was proposed to mimic the computation abilities of the plasmodium of Physarum polycephalum. This model was implemented on software, as well as, on hardware, namely on a Field Programmable Gate Array (FPGA). The specific ability of the P. polycephalum simulated here is given in brief, also bringing attention to the approximation of a Kolmogorov-Uspensky machine (KUM), an alternative to the Turing machine. KUM represent data and program by a labeled indirected graphs and a computation is performed by adding/removing nodes/edges. The proposed model implementation is taking full advantage of the inherent parallel nature of automaton networks, and CA, as a result of the mapping of the local rule to a digital circuit. Consequently, the acceleration of the computation for the hardware implementation, compared to the software, is as high as 6 orders of magnitude.