Toward the Tradeoffs between Privacy, Fairness and Utility in Federated Learning