Agent-Based and Mean-Field Models for: Multi-Stage Supply Chain Attacks

Nilesh D. Sadaphal

In digital supply chains, multi-stage cyber attacks slowly propagate, have long compromise states and are distributedly interdependent between operational entities. This study will provide a Fractional-Order Dynamical Framework for Multi-Stage Supply Chain Attack Propagation using the fractional diffusion dynamics and stochastic agent interaction and distributed recovery evolution. The modelling of memory dependent attack persistence is captured by fractional order derivatives, where the propagation behaviour of the attacks depends on the compromise states in history. Cascading attack dynamics between inter-connected stages of a supply chain are captured by including agent-level stochastic transitions and macro-level diffusion fields. Stability and resilience are assessed using persistence decay, recovery convergence and distributed risk evolution. Under memory-dependent dynamics, experimental evaluation showed that the attack diffusion error decreased to 0.104, the stage compromise rate to 22.6%, and the persistence index to 0.21 while the recovery stability index improved to 0.93, showing that the resilience of the attack was enhanced and that the attack propagation was controlled.