Terminal scheduling method and device in wireless federated learning

Embodiment of the invention provide a terminal scheduling method and equipment in wireless federated learning. According to the method provided by the embodiments of the invention, a scheduled terminal in each round of federated learning is dynamically adjusted through gradient information estimation carried out by each terminal in each round of federated learning and current round time consumption estimation carried out by a wireless access point, so model accuracy which can be obtained by federated learning can be maximized within the limited training time delay; and thus, the problem that the convergence speed of federated learning is difficult to guarantee under a dynamic wireless environment and variable training data distribution due to the fact that only certain preset fixed parameters can be used in an existing terminal scheduling algorithm is solved. 本发明实施例提供了无线联邦学习中的终端调度方法及设备，本发明实施例提供的方法，通过每轮联邦学习中各终端进行的梯度信息估计以及由无线接入点进行的当轮耗时估计，动态地调整每轮联邦学习中的被调度终端，可以在受限的训练延时内最大化联邦学习能够获得的模型正确率，从而解决了现有终端调度算法只能使用某些预设的固定参数