Some of the main difficulties in completing the “little brother” infrastructure analysis IoT Initiative include:
· Lack of available studies on critical physical infrastructure. Largely only bridges have undergone any analysis, and that analysis was limited to the state of Indiana.
· Lack of infrastructure deterioration models. The Markov models are a bit simplistic and primarily geared to visual inspections. The more consistent pattern matching and correlations with available public information (weather, etc.) should improve the models, but as with other aspects it will depend on the usage and will only improve over time.
· Lack of good modeling tools for the models needed for contextual analysis. The combination of IDENGINE and the Infrastructure Workbench will need to be integrated with the Pharo data collection and analysis tools.
· The models themselves will need to be constantly regenerated from the changing data. ACT-R is a good framework for dynamic regeneration of pattern matching and poisson based models, and to determine the quality of the heuristics used by the engineer themselves on specific problem types. The current pattern matching / deviation code is built in ACT-R for that reason. The ease of writing custom scripts and modifying the initial source manually will also help in improving the tooling.
· Average costs for repairing / replacing infrastructure are unavailable. By having infrastructure engineers enter the final cost a relatively accurate average can be determined over time. The larger and more geographically diverse the user base, the more accurate the estimates will become.
· The success of the initiative is highly dependent on the size and diversity of the system’s usage. The lack of tools currently available makes that usage far more likely, but assistance in making the hardware affordable will also increase usage, particularly in the targeted poorer areas of the world and of different areas of individual cities.