At&t Intellectual Property I, LP

At&t Mobility Ii LLC

Huijing Yang

Wenjie Zhao

Huahui Wang

Jie Chen

Ke Zhang

Slawomir Mikolaj Stawiarski

Ravi Raina

Gopalakrishnan Meempat

Weihua Ye

Hyun Ok Lee

Ganesh Krishnamurthi

Dongho Kim

Senthil Raj Dhandapani

Hendra Tuty

Kemal Kara

Jun Li

Yu Chen

A processing system including at least one processor may obtain operational data from a radio access network (RAN), format the operational data into state information and reward information for a reinforcement learning agent (RLA), processing the state information and the reward information via the RLA, where the RLA comprises a plurality of sub-agents, each comprising a respective neural network, each of the neural networks encoding a respective policy for selecting at least one setting of at least one parameter of the RAN to increase a respective predicted reward in accordance with the state information, and where each neural network is updated in accordance with the reward information. The processing system may further determine settings for parameters of the RAN via the RLA, where the RLA determines the settings in accordance with selections for the settings via the plurality of sub-agents, and apply the plurality of settings to the RAN.

Radio access network control with deep reinforcement learning

The disclosed technology is directed towards load balancing in an adaptive and automated way for wireless mobility networks to improve the overall harmonic-average UE throughput within each controlled group of cells (e.g., different frequency carriers serving a sector of a base station). A load balancer (e.g., analytics component) obtains various device traffic data including throughput data for cells of a group. Pairs of cells in a group (sharing a site and face) can be selected based on satisfying various criteria, with estimated throughput gain achieved by changing the handoff rates between the cell pairs. The technology iteratively repeats the overall process, driving a system to an optimal equilibrium.

Adaptive automated load balancing for mobility networks including fifth generation networks and beyond

A system includes a first network edge data collector, a first network edge key performance indicator (KPI) engine configured to operate on first data collected by the first network edge data collector, a KPI metrics manager in communication with the first network edge KPI engine, the KPI metrics manager controlling a KPI metric catalog and wherein the first network edge KPI engine determines first edge KPI metric using a metric algorithm from the KPI metric catalog on the first data.

System and method for a generic key performance indicator platform

A device can receive, from a network node device, call trace event data relating to characteristics of a wireless communication session between the network node device and a user equipment. The device can sequence and combine the call trace event data for a period of the wireless communication session. The device can analyze the call trace event data to determine a category of network communication traffic transmitted via a communication channel between the network node device and the user equipment. In response to a determination that the network communication traffic comprises streaming video packets, the device can facilitate directing of network resources to be allocated to support the wireless communication session.

Real-time user traffic classification in wireless networks

Growing community of inventors

Princeton, NJ, United States of America

Huijing Yang