Page 49 - AI Ready – Analysis Towards a Standardized Readiness Framework

AI Ready – Analysis Towards a Standardized Readiness Framework

Table 2: General use case analysis and AI impacts (continued)

Examples Potential AI impacts
Actors
7. Roaming across countries and regions for seamless connectivity [50]
8. Satellite networks [51] e.g. provide inputs on wildfire.
9. Ad hoc network design between drone clusters [52]
10. Mobile data network [2]
11. Model updates and data collection (for training and finetuning) over the
network [68]
12. Communication network reachable in remote regions [71]
13. Connectivity between edge and cloud models for distributed image
processing [1]
14. 5G/6G network operations, autonomy, and optimizations [93].
1. AI/ML feedback loop, collect data, infer, action recommendations and
policy application, pre-built traffic plan for specific occasions.
2. Opensource boards with ruggedization to IP65 standards.
3. Autonomous flight mechanisms for drones with image capture mecha-
nisms.
4. Continuous improvement of models using feedback and optimizations is
part of future work [50].
5. Fire detection, propagation models, and alarms to local communities and
utilities [51]
6. Closed loop: UAV network designed to autonomously perform essential
tasks within disaster-stricken areas [52]. UAVs can learn and adjust their
operations (including route navigation, returning to charging stations,
Experimen- and data detection and transmission) based on feedback from the envi-
tation and ronment.
controllers 7. Summary report generated and monitored via doctor dashboard [2]
with Sand- 8. Closed loop for Models which continuously learn from user feedback to
box-based enhance accuracy [59]
actors 9. Validation of TinyML models in AI Sandbox [67]
10. Finetuning general models with regional data in local Sandbox [68], [71]
11. Simulation environments such as Matlab/Simulink and SimPy [71] can
create experimentation abilities for various scenarios.
12. Experimentation and selection of the best model with a set of trials using
open-source libraries such as scikit-learn and XGBoost, with several runs
and for each trial run so you can review, reproduce, and modify the code
[77]
13. Cloud-based internal deployments for validation [80]
14. Cloud-hosted annotation sandbox [27] for international collaboration of
data annotation.
15. Digital twin of cloud resource management for simulation and optimiza-
tion in 5G/6G networks [93].

42