EDFS–MXD Demo Evaluation SDKGraph-Based Software Design Flow for Multi-Domain Systems
Graph-Based Software Design Flow for Multi-Domain Systems
Maxdi Inc. Cognitave Inc. Quantum Research Division
1 Overview
This technology note accompanies the EDFS–MXD Demo Evaluation SDK bundle delivered to
client teams for rapid evaluation. The SDK represents a graph-based Software Design Flow (SDF)
that complements existing EDA kernels (Keysight ADS, Cadence AWR, CST, FEKO, and open-
source KiCad) by adding a portable, versionable design-flow layer with executable stages, reporting
hooks, and model containers.
2 SDF Design Trajectory Exported from EDFS
Figure 1 shows a representative SDF instance exported from EDFS. Nodes represent validated
design stages; directed edges represent executable progression (data & intent flow), not merely connectivity.
Figure 1: Example Software Design Flow (SDF) exported from EDFS (PNG preferred; TikZ fall-back if absent).
3 MXD Disk Containers and NxS Analytics
In the full SDK, the SDF graph is paired with:
• MXD Disk behavioral containers (.cogn) representing deformed performance envelopes,
order-parameter evolution, and decision margins.
• NxS analytics that compute stability indicators, sensitivity surfaces, and integration risk
metrics over the SDF trajectory and exported lab/sim traces.
1Interoperable export formats. The SDK supports exports suitable for customer toolchains,
including .cogn containers, Touchstone-like network data, and data-cube structures (e.g., FMCW
radar baseband captures) that can be consumed by ADS/AWR data access features or open-source
processing stacks.
4 Evaluation Scope (TRL4/5)
This demo bundle is scoped for rapid evaluation and integration planning:
1. Import a graph, run the demo plot + report pipeline, and verify portability.
2. Validate that design-flow stages can bind to the customer’s existing simulation/test kernels.
3. Review the .cogn container concept for model distribution, versioning, and IP control.
5 Conclusion
EDFS–MXD introduces a design-flow operating layer that formalizes uncertainty, measurement
burden, and integration coupling as first-class objects. This enables more realistic assessment than
static margin stacking or heuristic-only expert judgment, particularly for multi-domain systems
operating under low margins.
2
