Closing the Unified Modeling Gap with EFSUT and EF-Builder
The defense modeling landscape is fragmented across transport, tracking, and kill-chain domains — each built on incompatible formalisms, temporal models, and data representations that no single tool has been able to bridge. The EFSUT methodology (Experimental Frame / System Under Test) directly addresses this structural gap by providing a rigorous, question-driven approach to cross-domain simulation design. Rather than forcing incompatible models into a single monolithic environment, EFSUT uses DEVS formal composition theory to define modular experimental frames — each scoped to a specific stakeholder question — that can observe and test system behavior across domain boundaries through explicit port-based coupling. The EF-Builder tool operationalizes this methodology, walking practitioners through a structured conceptual-to-developmental pipeline: from stakeholder question analysis and candidate frame generation, through formal verification of maximal experimental frames, to the production of deployable simulation code. The result is a repeatable, verifiable path from “what do we need to know?” to “run the experiment” — spanning transport kinematics, probabilistic tracking, and kill-chain decision logic within a single composable framework.
Developing Time Aware Decision Support Trading Platform
RTSync is currently developing a unified modeling and simulation infrastructure that integrates DEVS hierarchical modular construction, Hybrid DEVS, MBSE, and ParaDEVS into a coherent, operational framework for day and high‑frequency trading use-cases with reinforcement‑learning‑based portfolio optimization. This work extends and strengthens an already mature foundation in DEVS‑based modeling and simulation, bringing it directly into the domain of real‑time financial decision systems.
Our effort focuses on explicit, compositional, and auditable models that capture the hybrid nature of financial markets—continuous stochastic dynamics intertwined with discrete events, structural breaks, and regime shifts. We are leveraging the DEVS formalism to represent these systems with clarity and rigor, while MBSE provides the architectural discipline needed for traceability, reuse, and institutional adoption.
At the core of the system, we are implementing a multi‑resolution family of Ornstein–Uhlenbeck (OU)‑based hybrid DEVS models that generate time‑indexed volatility surfaces and factor trajectories. These models support real‑time scenario generation, stress testing, and risk‑aware decision support. ParaDEVS extends this capability by replacing traditional Monte Carlo sampling with structured uncertainty expansion, enabling it to explore full decision trees under varying conditions without losing auditability or interpretability.
In parallel, we are integrating reinforcement learning agents that operate on the outputs of the DEVS/ParaDEVS pipeline. These agents learn optimal trading and portfolio‑allocation strategies over time, using volatility‑surface geometry, tail‑risk metrics, and sensitivity measures (Greeks) as state features. The RL layer benefits from the DEVS infrastructure by receiving consistent, scenario‑rich, and model‑derived signals, rather than noisy or ad‑hoc market proxies.
Our Development Approach Includes
A modular DEVS/Hybrid DEVS library supporting continuous dynamics, discrete events, and their interactions.
A ParaDEVS engine for tree‑based uncertainty expansion with merging, pruning, and traceability.
A modeling notation and System Entity Structure ontology‑based configuration system enabling explicit model families, selection rules, and derivability.
A scenario‑generation and experimental‑frame layer that standardizes how stress tests, mission threads, and uncertainty regimes are defined and executed.
A transducer and analytics layer for extracting risk metrics, performance indicators, and decision‑relevant summaries.
A workflow and governance framework ensuring that every model instance, configuration, and experiment is reproducible, auditable, and suitable for institutional adoption
Enhance your modeling infrastructure with DEVS.
For high‑frequency trading applications, we are deploying the DEVS‑based models in a streaming environment where factor states, volatility surfaces, and risk metrics are updated continuously. The system supports time-aware decision cycles, allowing coarse‑resolution models to deliver early guidance while higher‑resolution models refine the picture as simulation experiments proceed. This progressive‑resolution approach ensures that trading decisions remain grounded in the best available evidence at every moment. This positions the infrastructure as a robust operational capability for real‑time financial decision making.
