A dual-polarization paralleled carrier phase recovery algorithm with better linewidth tolerance and lower computational complexity is presented for polarization division multiplexing and 16-order quadrature amplitude modulation (PDM-16QAM) system based on linear Kalman filter (LKF). The symbols in first and third rings of two polarizations are used to perform phase noise estimation. The simulation results for PDM-16QAM system with the transmission rate of 224 Gb/s demonstrate that the linewidth tolerance of dual-polarization LKF is about seven times of that of single-polarization LKF. The linewidth tolerance is improved to 2800 kHz from 400 kHz. The dual-polarization LKF processes about four times symbols in one block than single-polarization LKF, which reduces the algorithm computational complexity and enhances the real-time performance. Finally, the proposed algorithm is verified in the transmission experiment of PDM-16QAM system with the transmission rate of 224 Gb/s.