Abstract: One example method is a method for improving O-RAN (open radio access network) LLM (large language model) performance. The example method includes monitoring performance of a task in an O-RAN by an LLM running at an agent deployed in the O-RAN, receiving information from the agent concerning performance of the task, based on the information, generating an update to the LLM, and transmitting the update to the agent, and the update is usable by the agent to update the LLM and/or the associated KG.

Abstract: One example method includes generating an enhanced network graph (ENG) that is a representation of a physical RAN, providing the ENG as input to a GNN model, with the GNN model, using the ENG, to obtain network state embeddings of the physical RAN, providing the network state embeddings and a cluster of network task requests to a prompt generator, by the prompt generator, creating a prompt based on the network state embeddings and network task requests, transmitting the prompt to an AFM (agentic foundation model), based on the prompt, capturing, by the AFM, entity interdependencies of the physical RAN, transmitting the prompt, by the AFM, to a cluster of downstream tasks that each correspond to a respective one of the network task requests, and by the cluster of downstream tasks, generating respective control commands and transmitting the control commands to the physical RAN.

Abstract: Generalized network incident management is disclosed. When an incident in a network is detected, a request is generated and sent to a prompt generator. The prompt generator generates a prompt based on the request, key performance indicators of the network, and a semantic state of the network, which is generated by a first model. The prompt is input to a second model and the second model generates a recommended solution to the incident. An agent may evaluate the recommended solution and send a command to the network to resolve the incident. The prompt generator also generates the prompt using a graph neural network and/or a knowledge base. The network incident management can be automated and/or include a human-in-the-loop.

Abstract: One example method operates to configure and optimize a private communication network, and includes receiving, by a prompt generator, input comprising a KG (knowledge graph), and/or a base network configuration generated by a first LLM (large language model), using, by the prompt generator, the input to create an initial network design prompt, receiving, by a second LLM, the initial network design prompt, and using, by the second LLM, the initial network design prompt to create a site-specific network configuration, providing the site-specific network configuration to a network orchestrator, and orchestrating, by the network orchestrator, the site-specific network configuration to deployment site.

Abstract: One example method includes obtaining data concerning a configuration of a network, and passing the data to a KG (knowledge graph) creation engine, by the KG creation engine, using a KG creation LLM (large language model) to extract knowledge from the data, generating, by the KG creation LLM, a portion of a KG that represents a portion of the network, by a judge LLM, evaluating the portion of the KG to determine an extent to which the portion of the KG accurately represents the portion of the network, and updating a ground truth KG with the extracted knowledge when the portion of the KG has a confidence score that exceeds a threshold confidence score.

Abstract: Generalized risk management is disclosed. When an event related to a network is identified, a request is generated and sent to a prompt generator. The prompt generator generates a prompt based on the request, key performance indicators of the network, and a semantic state of the network, which is generated by a first model. The prompt is input to a second model and the second model identifies risks associated with the event and/or generates a recommended potential solution to a potential incident caused by the event should the event occur. An agent may evaluate the risks and/or the recommended potential solution and send a command to the network to mitigate the risks and/or resolve the incident should the incident occur. The prompt generator also generates the prompt using a graph neural network and/or a knowledge base. The network incident management can be automated and/or include a human-in-the-loop.

Abstract: Network data management that includes data compression is disclosed. Multi-modal data from a network is collected and synchronized. A multi-modal model is trained to select data, in response to a prompt from a task model, based on the request, the multi-modal data, and/or data retrieved from a knowledge graph of the network. The selected multi-modal data is compressed and/or cleaned by the multi-modal model and returned to the task model. The task model is configured to perform operations or requests from applications of a radio intelligent controller associated with a network.

Abstract: A model-based network digital twin for forecasting and scenario generation is disclosed. The network digital twin includes twin components that each include a component model and a component knowledge graph. Each twin component is associated with a corresponding network component. Data from the network is ingested into the network digital twin and used to update the knowledge graphs. Requests received through an orchestration model are input to the component models, which generate responses using the corresponding component knowledge graphs. The orchestration model generates a final response to the request for an operator. The request may be a forecast or what-if scenario.

Abstract: Foundational model tuning for network management and application is disclosed. Network feedback is generated by a physical network (RAN), a network model included in a digital twin or is received from other sources, including a network operator, and stored in a database. The network feedback stored in the database, along with graph data from a knowledge graph of the network, is used to generate a prompt and train a model copy using reinforcement learning. Updates derived or obtained from the model copy are incorporated into the network model. This allows the network model to be tuned to the network and network management and monitoring operations.

Abstract: Enhanced network graphs and graph neural networks (GNNs) are disclosed. In a system that includes a radio access network and a radio intelligent controller (RIC), the RIC is configured to receive multi-modal data along with other data. Semantic features from the multi-modal data are combined with features from the other data. The combined features are used to enhance a network graph of the network and to train a GNN whose feature embeddings are enhanced with the multi-modal data. Decisions made in the network may be based on the feature embeddings, which account for multi-modal data and network inter-dependencies.

Abstract: Policy generation and/or implementation in networks including open radio access networks (O-RAN) is disclosed. A radio intelligent controller (RIC) is configured with models to facilitate automatic policy generation. In one example, a policy intent is mapped to an imperative objective by a first model. Data retrieved from a knowledge graph, which knowledge graph is updated by a second model, is augmented with the imperative objective to generate a prompt. A third model, which also received the network's telemetry data, generates a recommended policy for implementation in the network.

Abstract: The technology disclosed herein enables worker nodes to reestablish connections when the connection protocols fail to successfully reconnect on their own. In a particular example, a method includes establishing a plurality of TCP-based connections (e.g., iSCSI and/or NVMe) between a plurality of worker nodes and a storage system and determining status of the plurality of TCP-based connections. When the status indicates a first connection of the plurality of TCP-based connections has failed, the method includes determining credentials used to establish the first connection are invalid. In response to determining the credentials are invalid, the method includes requesting new credentials from a controller node and reestablishing the first connection using the new credentials.

Abstract: A switching amplifier circuit includes an output transistor, a current direction detection circuit, and a slew control circuit. The output transistor includes a control terminal coupled to a switching input signal; a drain node coupled to an output node coupled to a first end of a load device having the first end and a second end; and a source node coupled to a reference voltage. The current direction detection circuit is coupled to the output node and configured to detect a direction signal corresponding to an output current at the output node. The slew control circuit is configured to adjust a slew rate of the output transistor at the output node. The slew control circuit is coupled to the control terminal of the output transistor and is activated only during turnoff of the output transistor and operates in response to the direction signal.

Abstract: An output driver with slew rate control includes an output transistor that includes a control terminal coupled to a switching input signal, a drain node coupled to the output node for coupling to a load device, and a source node coupled to a reference voltage. The output driver also has a slew control circuit including a current source coupled in series at a connection node with parallelly connected first switch transistor and second switch transistor. The connection node is coupled to the control terminal of the output transistor. The first switch transistor has a control terminal coupled to the switching input signal. The second switch transistor has a control terminal that is coupled to either the switching input signal or a dynamically modulated switching input signal, depending on a current direction at the output node.

Abstract: In various embodiments, a computer-based loan decisioning system includes a data ingestion engine programmed for automatically ingesting information associated with making a financial decision for a borrower, and for receiving financial data associated with the borrower from a variety of file formats and from multiple external databases. A data extraction module may be used in the system for automatically parsing and classifying the ingested information and received data. Also, an analytic engine can be provided for performing calculations and financial analysis in response to the parsed and classified information and data. The analytic engine can assist with making financial decisions associated with the borrower.

Abstract: An output driver with slew rate control includes an output transistor that includes a control terminal coupled to a switching input signal, a drain node coupled to the output node for coupling to a load device, and a source node coupled to a reference voltage. The output driver also has a slew control circuit including a current source coupled in series at a connection node with parallelly connected first switch transistor and second switch transistor. The connection node is coupled to the control terminal of the output transistor. The first switch transistor has a control terminal coupled to the switching input signal. The second switch transistor has a control terminal that is coupled to either the switching input signal or a dynamically modulated switching input signal, depending on a current direction at the output node.

Abstract: A path generating device is configured to generate a path for a robot formed by connecting a plurality of units that are each bendable to have a desired single curvature, and the path generating device includes: an analysis unit configured to output position posture information indicating a position and a posture of the robot corresponding to an operation amount, using a robot model with which the position and the posture are able to be simulated in a virtual space; a generating unit configured to generate a path extending from a predetermined entry position to a target position in the virtual space; and a specification unit configured to specify an operation amount for making the robot model advance along the path, while making a position of a connection portion of each of the units of the robot model match the path.

Abstract: A computer-implemented system and method for associating immune system repertoires with specific stimuli (exposures) based on the biophysical properties of the repertoire's receptors. Sequences of a training repertoire are converted into a set of biophysical properties, and a computer-based compact representation of the training repertoire is built using maximum entropy modeling. In one version, an “immunome-wide association study” is performed by computer scoring a test repertoire using several such models to classify the test repertoire as being associated with a biological condition or not. In another version, one or more sets of parameters from the models are found that together classify each model as being from an individual that has the condition or from an individual that does not.

Abstract: The present disclosure provides systems and methods for generating routing assignments. One example method includes obtaining delivery data associated with a plurality of objects to be delivered from an object distribution hub to a plurality of delivery destinations, wherein a plurality of loading areas are associated with the object distribution hub; obtaining data indicative of a current location for each of the plurality of objects within the object distribution hub; determining data indicative of a respective amount of time required to transfer each of the plurality of objects from its current location to each of the plurality of loading areas; determining routing assignment data based at least in part on the delivery destinations and the respective amount of time required to transfer each of the plurality of objects to each loading area; and providing the routing assignment data to facilitate loading and delivery of the set of one or more objects.

Abstract: Various embodiments are generally directed an apparatus and method for determining a profile for an application, the profile to specify a setting for one or more storage services provided by a storage system, determining whether settings for provided storage services for the application conform to the profile. Further and in response to determining one or more of the provided storage services is non-conforming, performing a remediation operation to correct non-conforming storage services, and in response to determining the provided storage services are conforming storage services, providing an indication indicating the provided storage services are conforming to the profile.