Modeling & Simulation

Fidelity

FavoriteLoadingAdd to favorites

 

Fidelity is the accuracy of the model or simulation when compared to the real world. Simulation fidelity has to do with how well the simulation responds and how the results correspond to what the simulation is trying to represent. Fidelity is important because it is what distinguishes models and simulations from other computer programs because at its core it represents a simuland; a system being simulated by a simulation. [1]

It is important to distinguish between what the simulation is intended to represent (the simuland), and what it is actually able to represent.  The simuland is often casually referred to as the “real world” or as reality, actuality, or truth.  However, simulands are not necessarily the “real world,” because many simulations do not intend to represent situations found in current reality.  For example, they may represent the performance of proposed weapon systems in a hypothetical battlefield. [1]

Because any model or simulation is, by definition, an abstraction or representation of some part of reality (material or imagined), fidelity is interwoven with all facets of model and simulation development. Below is a number of recommended practices when address fidelity. [1]

  • Recognize that fidelity is a core concept spanning every issue in simulation, especially issues related to V&V. The distinguishing characteristic of simulations is that they are systems that contain within themselves a model of another system. Fidelity is at the core of understanding how to specify the representational requirements and validate that the requirements and eventual model suitably represent that “other system”. Users, M&S Program Managers (PM), Developers, V&V Agents, or Accreditation Agents can and should think about fidelity impacting their projects in terms of the fidelity framework.
  • Beware single point or qualitative fidelity descriptions. While fidelity is a unified concept, it has little or no meaning when expressed as a single point or qualitative description (e.g., low, medium, or high). Simulation fidelity can and should be decomposed into its constituent components of resolution, error/accuracy, sensitivity, precision, and capability. When presented with single point or qualitative fidelity descriptions, a User, M&S PM, Developer, V&V Agent, or Accreditation Agent should seek meaningful insights by asking about the model’s resolution, error/accuracy, sensitivity, precision, and capability. Likewise, they all should push toward specificity in representational requirements, which will inevitably address resolution, error/accuracy, sensitivity, precision, and capability needed — as opposed to requiring “goodness”.
  • Use comparison as a basis for defining the fidelity aspects of representational requirements. Without resorting to the various quantitative methods being proposed in the research community, in practice, the fidelity of the proposed simulation can be compared to simulations meeting similar purposes in order to gauge its fitness for purpose
  • Seek to limit the fidelity required and implemented to that which is actually needed. Frequently simulations projects seek to include all the fidelity they can afford, without realizing the burden that creates and the reduced benefit that results. Higher fidelity simulations cost more time and money to build, more to V&V, and more to operate. Furthermore, the perceived increase in quality with higher fidelity is sometimes illusionary

See the Verification, Validation & Accreditation Recommended Practice Guide for more detailed information about fidelity.


AcqTips:
 

  • Fidelity can be a complex term used in the Modeling and Simulation community. Make sure you a clearly defined definition of fidelity before developing, updating or running any simulations.

AcqLinks and References: 

Become an AcqNotes Member to View Page Discussions