CoSimu

The CoSimulator to understand cyber-physical consequences.

State exchange Injected disturbance Injection point Clock sweep
t+00:04:12.4 · 6/6 models coupled

CoSimu is developing reusable twin kernels of physical processes and reusable security-analysis libraries, so that a cyber intervention can be examined across interconnected physical processes within a controlled experimental environment.

The platform Research directions

Investigating cyber-physical consequences in interconnected systems

Cyber interventions can alter the measurements, commands and control decisions that govern physical processes. What follows depends on the dynamics of the affected system, its operating conditions, and the processes coupled to it.

Detecting an intervention and understanding its physical effect belong together, and both inform the evaluation of a defensive measure. Studying the relationship between them requires an experimental environment in which cyber and physical behavior are represented within one system context.

CoSimu is being developed to provide such an environment.

Reusable infrastructure for cyber-physical security research

Cyber-physical security experiments are typically constructed for one system and cannot be carried into the next study. CoSimu separates the representation of a physical process, and the method used to investigate it, from the machinery that couples them for a particular run.

The CoSimu architecture Process-specific twin kernels on the left and reusable security-analysis libraries on the right are selected per experiment and composed by an orchestration layer in the center, which couples the chosen models, exchanges state across their declared interfaces, holds a common time base and applies controlled interventions. The composed experiment produces evidence along a timeline running from cyber intervention through control response and physical consequence to analysis. Twin kernels reusable representations of physical processes Security-analysis libraries reusable experimental and analytical methods Power systems EMT · phasor · protection logic Industrial processes control loops · actuation Water distribution hydraulics · pumping · storage Further systems manufacturing · transport Attack scenarios measurement · command manipulation Physics-informed detection contextual anomaly analysis Moving target defense sensitivity-guided perturbation Disturbances & faults operational and equipment events Orchestration Model coupling selected per experiment State exchange across declared interfaces Common time base negotiated exchange step composed experiment Experimental evidence Cyber intervention forged measurement Control response regulator acts on false state Physical consequence process leaves its envelope Analysis observability · margin · comparison
Twin kernels

Reusable representations of physical processes.

Power systems
EMT · phasor · protection logic
Industrial processes
control loops · actuation
Water distribution
hydraulics · pumping · storage
Further systems
manufacturing · transport
Security-analysis libraries

Reusable experimental and analytical methods.

Attack scenarios
measurement · command manipulation
Physics-informed detection
contextual anomaly analysis
Moving target defense
sensitivity-guided perturbation
Disturbances and faults
operational and equipment events
Orchestration
Model coupling
selected per experiment
State exchange
across declared interfaces
Common time base
negotiated exchange step
Experimental evidence
01
Cyber intervention
forged measurement
02
Control response
regulator acts on false state
03
Physical consequence
process leaves its envelope
04
Analysis
observability · margin · comparison

Development status is set out on the Platform page.

Twin kernels

Reusable computational representations of physical processes, exposing the dynamics, state variables, parameters, interfaces and operational constraints that other components need in order to interact with them.

Security-analysis libraries

Attack scenarios, disturbances, detection methods and defensive strategies expressed against declared interfaces, so that a methodology can be applied to further systems whose dynamics are compatible.

Orchestration

Binds the selected kernels, transfers state across their interfaces, coordinates how they advance in time, and applies interventions at defined points in the experiment.

Experimental evidence

A recorded account of how an intervention propagated, through which interfaces, against which physical constraints, and what could be observed from a given monitoring position.

The portability of a security methodology depends on the physical dynamics, interfaces and operational assumptions of the system to which it is applied. A detection method developed against power-system dynamics is re-expressed and re-validated for a hydraulic process: the methodology carries across, while the configuration and its validation stay specific to the system. The architecture in detail →

From cyber intervention to physical response

A coupled experiment examines how a manipulated measurement or control command affects behavior across interconnected processes.

Propagation and observability across one coupled run Two stacked schematic traces over a shared time axis. In the upper lane a process variable holds inside its operating envelope until a measurement is manipulated at t0; the regulator acts on the false state at t1 and the variable leaves the envelope at t2. A second trace shows the same composition re-run with a defensive configuration active, in which the variable stays inside the envelope. In the lower lane a conventional monitor's residual stays flat throughout, while a physics-informed residual rises and crosses the alarm threshold shortly after the controller responds. PROCESS VARIABLE operating envelope intervention applied, no additional defense same configuration with a defense active DETECTOR RESIDUAL alarm threshold network monitor: packets well formed, values within range physics-informed residual exceeds the threshold t₀ t₁ t₂ t₃ measurement manipulated controller acts envelope exceeded evidence recorded
Baseline configuration With a defense active Network monitor
Cyber intervention, control response and physical trajectory recorded along a common experimental timeline.
t₀
A measurement is manipulated
The value remains plausible and the packets remain well formed, so a network monitor registers nothing unusual.
t₁
The controller responds
The regulator acts on a condition that does not exist, and the intervention passes from the communication path into the physical process.
t₂
The process leaves its operating envelope
Physical behavior departs from the operating band, and the departure reaches the processes coupled to it.
t₃
The experiment is recorded
The record covers the propagation path, the physical constraints that shaped the response, and what each monitoring position was able to observe.
↻
The configuration is compared
Repeating the experiment with a single element altered isolates the effect of that element on both the physical trajectory and the available evidence.

Coupling the models produces a record that crosses domain boundaries: the same event as observed from the communication path, from the control system and from the physical process, along one experimental timeline. Because everything other than the element under study is held constant, configurations can be compared directly.

The comparison shows whether a defensive measure changed the physical trajectory, or only the evidence available to the defender. The strength of either conclusion depends on the fidelity of the models, the interfaces selected and the design of the experiment.

Research directions

Four connected lines of inquiry shape the architecture, and each determines what an experiment built on it can be designed to answer.

Heterogeneous model coupling

The conditions under which models with different time representations, state variables and modeling assumptions can participate in one experiment while the resulting evidence remains valid.

Attack propagation and observability

The pathway an intervention takes through measurement, control and physical behavior, and the evidence that pathway leaves at different monitoring positions.

Physics-informed security

How process dynamics and operational constraints can contribute to detection and to defensive design, and what such methods cost in operation.

Cyber-physical resilience

How far a disturbance propagates, how severely operation is affected, how long degradation persists, and whether a defensive configuration alters the physical trajectory.

Research program

CoSimu is under active development. Inquiries concerning cyber-physical security research, experimental models, security-analysis methodologies and potential research collaborations are welcome.

Platform Research FAQs Contact
Nouman Nafees, PhD
Principal Investigator