the AI Planner
The AI planner, working with the mission rApp, addresses the communication needs of military or civilian personnel during missions or emergency scenarios. It does this by creating 5G network slices with specific coverage and performance characteristics.
AI planning is a branch of Artificial Intelligence that uses existing expert knowledge to solve problems rather than learning from examples. In networking, this concept is similar to “intent-based networking,” where AI planning is applied to manage network issues.
In emergency scenarios, multiple operators each have their own 5G network, covering different areas. Each operator has its own Service Management Orchestrator (SMO), and there’s a higher-level coordinating SMO (MO-SMO) that the AI planner communicates with.
The AI planner creates network slices in this multi-operator environment. During disasters or military actions, it coordinates resource-sharing among operators through the MO-SMO. This requires that networks have a disaggregated structure, allowing the MO-SMO to see and use resources from different networks.
The main needs in these scenarios are quick decision-making and explanations that humans can understand and act on. Our hierarchical task network planner satisfies these needs better than machine-learning approaches. This planner acts like a manager, breaking down tasks into simpler sub-tasks that can be handled by operator SMOs.
The planner searches for the best way to move from an initial state to a desired outcome, using expert methods to make the search efficient. It is based on the SHOP3 HTN platform, which allows complex computations to evaluate conditions and determine actions.
Currently, we are developing coverage and provisioning algorithms for the planner. These algorithms are chosen based on the current state of networks and user preferences, ensuring optimal performance. We also have a prototype of the planner implemented as a literate program, which is transparent and easy to understand.
AI planning is a branch of Artificial Intelligence that uses existing expert knowledge to solve problems rather than learning from examples. In networking, this concept is similar to “intent-based networking,” where AI planning is applied to manage network issues.
In emergency scenarios, multiple operators each have their own 5G network, covering different areas. Each operator has its own Service Management Orchestrator (SMO), and there’s a higher-level coordinating SMO (MO-SMO) that the AI planner communicates with.
The AI planner creates network slices in this multi-operator environment. During disasters or military actions, it coordinates resource-sharing among operators through the MO-SMO. This requires that networks have a disaggregated structure, allowing the MO-SMO to see and use resources from different networks.
The main needs in these scenarios are quick decision-making and explanations that humans can understand and act on. Our hierarchical task network planner satisfies these needs better than machine-learning approaches. This planner acts like a manager, breaking down tasks into simpler sub-tasks that can be handled by operator SMOs.
The planner searches for the best way to move from an initial state to a desired outcome, using expert methods to make the search efficient. It is based on the SHOP3 HTN platform, which allows complex computations to evaluate conditions and determine actions.
Currently, we are developing coverage and provisioning algorithms for the planner. These algorithms are chosen based on the current state of networks and user preferences, ensuring optimal performance. We also have a prototype of the planner implemented as a literate program, which is transparent and easy to understand.
