Abstract: Systems and methods are provided for training Reinforcement Learning (RL) policies for a number of agents distributed throughout a network. According to one implementation, a method associated with an individual agent includes participating in a training process involving each of the multiple agents, the training process including multiple rounds of training allowing each agent to perform a local improvement procedure using RL. During each round of training, the method also includes performing the local improvement procedure using training data associated with one or more other agents having a relatively high level of affiliation of different levels of affiliation with the individual agent and additional training data associated with the individual agent itself. According to additional embodiments, a controller may coordinate the local improvement procedures. During inference, the RL policies can be used without the help of the controller.

Abstract: In various embodiments, the present disclosure relates to systems and methods for enhanced reinforcement learning (RL) algorithms using future state prediction. In some embodiments, an offline emulator can be applied allowing the generation of samples, thus supporting continuous training of the system and fast-forward fabric saturation. The fabric accepts transactions which allocate resources with respect to the transactions needs and constraints and contains an RL/AI model(s) which are continuously learning based on the current reward combined with reward scaling. By modelling the fabric and transactions in an emulator, it is possible to predict future states and calculate adjusted rewards with respect to the optimal criterion. A state generator is based on modeling past historical transactions, allowing a user to anticipate future state characteristics of the fabric. In some embodiments, online learning is based on adjusted rewards which are more representative with respect to the objective function.

Abstract: In various embodiments, the present disclosure relates to systems and methods for enhanced reinforcement learning (RL) algorithms using future state prediction. In some embodiments, an offline emulator can be applied allowing the generation of samples, thus supporting continuous training of the system and fast-forward fabric saturation. The fabric accepts transactions which allocate resources with respect to the transactions needs and constraints and contains an RL/AI model(s) which are continuously learning based on the current reward combined with reward scaling. By modelling the fabric and transactions in an emulator, it is possible to predict future states and calculate adjusted rewards with respect to the optimal criterion. A state generator is based on modeling past historical transactions, allowing a user to anticipate future state characteristics of the fabric. In some embodiments, online learning is based on adjusted rewards which are more representative with respect to the objective function.

Abstract: A node in a self-assembling network includes one or more radios connected to any of other nodes in the self-assembling network and user equipment, wherein at least one of the node and the other nodes is connected to a fiber or satellite for backhaul; and one or more processors and memory storing control software, wherein the control software is executed in a distributed manner with the other nodes, and wherein the control software is configured to monitor any of positions of the nodes, monitor bandwidth of user equipment connected to any of the nodes, and monitor quality of service requirement of the user equipment, and cause performance of adjustments to the nodes based on any of the monitored positions, the monitored bandwidth, and the monitored quality of service.

Abstract: A controller associated with a domain includes a network interface; one or more processors communicatively coupled to the network interface; and memory storing instructions that, when executed, cause the one or more processors to communicate with one or more additional controllers via the network interface, wherein each of the one or more additional controllers is in one or more additional domains, and wherein each domain provides different characteristics, utilize at least part of a control pattern to obtain requirements for a service, and cause, utilizing any of a peer relationship and a hierarchical relationship with the one or more additional controllers, at least part of implementation of a composition of resources to meet the requirements for the service, wherein the composition defines the resources provided in each domain for the service, and wherein the composition is based on the requirements and the different characteristics in each domain.

Abstract: A controller associated with a domain includes a network interface; one or more processors communicatively coupled to the network interface; and memory storing instructions that, when executed, cause the one or more processors to communicate with one or more additional controllers via the network interface, wherein each of the one or more additional controllers is in one or more additional domains, and wherein each domain provides different characteristics, utilize at least part of a control pattern to obtain requirements for a service, and cause, utilizing any of a peer relationship and a hierarchical relationship with the one or more additional controllers, at least part of implementation of a composition of resources to meet the requirements for the service, wherein the composition defines the resources provided in each domain for the service, and wherein the composition is based on the requirements and the different characteristics in each domain.

Abstract: A bandwidth control method implemented in a Software Defined Networking (SDN) network includes obtaining data for one or more services in the network, wherein each of the one or more services is controlled by an associated user-agent; determining future bandwidth requirements for the one or more services based on the data; determining service requests for at least one of the one or more services based on the future bandwidth requirements; and causing implementation of at least one of the service requests in order of priority. The process of prioritization uses a programmable network-wide logic and has the ability to consider information external to the network such as a user's Service Layer Agreement (SLA) and business priority. The entire bandwidth control method can repeat in cycles providing near real-time adjustments.

Abstract: Path calculation systems and methods for determining a path, based on constraints and rules, for a connection at one or more layers in a network, include determining a path exploration map of the network based on a multi-layer network model of the network defining external edges and intra-node paths, the path exploration map comprising every external port in the network that is reachable from a source port; and utilizing the path exploration map to determine a viable path, from a destination port to the source port, subject to the constraints and the rules and based on a cost.

Abstract: A bandwidth control method implemented in a Software Defined Networking (SDN) network includes obtaining data for one or more services in the network, wherein each of the one or more services is controlled by an associated user-agent; determining future bandwidth requirements for the one or more services based on the data; determining service requests for at least one of the one or more services based on the future bandwidth requirements; and causing implementation of at least one of the service requests in order of priority. The process of prioritization uses a programmable network-wide logic and has the ability to consider information external to the network such as a user's Service Layer Agreement (SLA) and business priority. The entire bandwidth control method can repeat in cycles providing near real-time adjustments.

Abstract: Path calculation systems and methods for determining a path, based on constraints and rules, for a connection at one or more layers in a network, include determining a path exploration map of the network based on a multi-layer network model of the network defining external edges and intra-node paths, the path exploration map comprising every external port in the network that is reachable from a source port; and utilizing the path exploration map to determine a viable path, from a destination port to the source port, subject to the constraints and the rules and based on a cost.