Towards Node Liability in Federated Learning: Computational Cost and Network Overhead