Model-Agnostic Semantics in Shipboard Systems Against Data Availability Attacks

Data availability attacks, such as latency attacks, data dropouts and time synchronization attacks (TSAs) still remain a prime concern not only at the network level but also impair the control system performance & stability of the physical layer in medium voltage DC (MVDC) shipboard power systems (SPSs). To address its impact on the physical layer, we equip a model-agnostic semantic architecture that can compensate using process-aware delay compensation capabilities. Unlike traditional model predictive controllers (MPCs) with a limited prediction horizon, the proposed architecture offers long event-driven prediction even during large random delayed measurements. The semantic prediction policy is governed using the inner control loop dynamics of power generation modules (PGMs) to provide reconstructed signals for delay compensation. Its robustness has been extensively tested and validated on nominal 12 kV two-zone MVDC SPS, in an OPAL-RT environment for above-mentioned attacks. Overall, the proposed model-agnostic estimator (MAE) has the potential to significantly improve the resilience of SPSs against cyber-attacks without revealing any model information.