Abstract

We study the dynamics of coupled oscillator systems in reconfigurable electronic circuits using field-programmable gate arrays (FPGA). These systems utilize pulse-width processing structures to realize individual chaotic oscillators as one-dimensional iterated maps with pulse-width state variables. However, real-world effects impose deviations from ideal models due to history effects that are not well understood or characterized. These effects can cause pulse width expansion and pulse-timing equalization in ring oscillator circuits. For networks, these behaviors provide coupling mechanisms which we exploit in sequential arrays of pulse-width oscillators. The resulting hardware arrays iterate using the same physical circuit paths thus minimizing electronic design mismatch between oscillators. We analyze the resulting network behavior using mutual information to detect coupling in quantized symbol streams generated by the multiplexed system. This analysis is motivated by the need to provide unique signatures and true randomness backed by first principles theory for cybersecurity applications.