Ctrl-P to Print

Complex Systems Architectures

Initiative Lead: Dr. Robert Leheny, Acting Director

Many natural (“non-engineered”) systems can be regarded as complex systems, including the weather, celestial mechanics and biological systems. These complex systems tend to have many parts that are highly interconnected through nonlinear interactions, and the influence of these parts on performance may be highly dependent on the states of other parts of the system. Further, the state-space for such systems may be so large as to defy clear definition and comprehensive examination by current techniques.

An example of a complex system is found in modern semiconductor circuits where processing techniques are permitting the realization of integrated microsystems with extremely high device density and increasing conditional interactions. As scaling trends continue to reduce critical dimensions and increase total component counts, the challenges associated with understanding and modeling the performance of such integrated microsystems become more significant. Traditional engineering has addressed this issue by breaking systems down into manageable subsystems, relying on the principal of superposition, whereby the response of each part can be treated separately and the system becomes the “sum of its parts.” Further simplification has been achieved by utilizing “approximately linear” models of real components and subsystems. These approaches come with a price, in that performance is often sacrificed in favor of realizing “practical” systems that exhibit a high degree of predictability. Moving forward, this engineering perspective clearly becomes untenable, as performance requirements and the push for integration continue to become more demanding while large engineered systems cross into the realm of genuinely “complex systems” and begin to exhibit unanticipated performance and features. Such observations are also generally true beyond the realm of integrated microsystems, including the engineering of macroscopic systems such as the internet.

The CSA effort seeks to exploit how an understanding of complex systems and in particular, the architecture of complex systems can be applied to benefit a broad array of U.S. Department of Defense (DoD) problems and the larger US national interests. CSA aims to apply Complex Systems Architectures science to the development of systems with enhanced capability over that currently achievable, while simultaneously improving operational robustness and functionality in the event of partial system failure. In pursuit of these aims, CSA would like to formalize design, development and testing methodologies that can concurrently shorten system development cycles while increasing overall confidence of system performance.