Incentive mechanism design for resource sharing in collaborative edge learning

In 5G and Beyond networks, Artificial Intelligenceapplications are expected to be increasingly ubiquitous. Thisnecessitates a paradigm shift from the current cloud-centricmodel training approach to the Edge Computing based collaborative learning scheme known as edge learning, in whichmodel training is executed at the edge of the network. In thisarticle, we first introduce the principles and technologies ofcollaborative edge learning. Then, we establish that a successful,scalable implementation of edge learning requires the communication, caching, computation, and learning resources (3C-L)of end devices and edge servers to be leveraged jointly in anefficient manner. However, users may not consent to contributetheir resources without receiving adequate compensation. Inconsideration of the heterogeneity of edge nodes, e.g., in termsof available computation resources, we discuss the challengesof incentive mechanism design to facilitate resource sharing foredge learning. Furthermore, we present a case study involvingoptimal auction design using Deep Learning to price fresh datacontributed for edge learning. The performance evaluation showsthe revenue maximizing properties of our proposed auction overthe benchmark schemes.