Energy Efficient Wireless Crowd Labeling Joint Annotator Clustering and Power Control

The unprecedented growth of mobile data traffic has fueled the deployment of intelligence applications at the network edge, while distilling the intelligence from raw data by supervised learning requires tremendous labelling effort. To overcome this challenge, wireless crowd labelling (WCL) is proposed for efficient data labelling by exploiting billions of available mobile annotators and the multicasting property of wireless channels. A WCL system is considered in this paper where unlabelled data (objects) are multicast to different clusters of annotators for repeated labelling to improve accuracy. Given the number of objects to be labelled with the required accuracy, the annotator clustering and multicast power control are jointly optimized for energy consumption minimization, leading to a challenging combinatorial problem. An efficient algorithm is designed to derive the optimal solutions based on tree search and examined by the simulations.