Artificial intelligence machine learning (AIML) inference offloading

Abstract

This disclosure provides methods, components, devices and systems for artificial intelligence and machine learning (AIML) inference offloading. Some aspects more specifically relate to techniques in which a first wireless node (such as an access point (AP)) may output one or more first messages indicative of a capability of the first wireless node to support AIML inferencing on behalf of a second wireless node (such as a station (STA). The first wireless node may then obtain, from the second wireless node, a request frame that requests the AIML inferencing by execution of at least one AIML model. In some aspects, the request frame may include at least one input vector associated with the at least one AIML model. The first wireless node may execute the at least one AIML model using the at least one input vector and output a response frame that includes an inference to the second wireless node.

Images