Early Detection of Cyberattacks in Banking Networks via a Fractional Partial Differential Equation Model

In this paper, we model edge traffic with a conformable fractional partial differential equation that keeps memory in time and space. The solution represents a unit‐free attack pressure, built from a z‐scored edge series, a quiet period baseline, and a partially absorbing boundary that reflects scrubbing and rate limits. Using a closed form atomic solution, we split the edge signal into a growth part and a damped part for early warning, and use a simple spatial factor to estimate how far pressure can travel inside. The result is a small, readable model that turns one edge metric into clear, tier‐specific actions during DDoS attacks.