Pilot-Assisted Maximum-Likelihood Estimation for Underwater Acoustic Communication

This paper addresses the problem of coherent detection of orthogonal frequency division multiplexing (OFDM) signals transmitted for underwater communication (UWC). The channel output is distorted by carrier phase and frequency offset, besides complex white Gaussian noise. A frame structure with a known preamble, a cyclic prefix (CP), followed by data is proposed for OFDM UWC. Maximum likelihood (ML) estimation is used for pilot symbol-assisted modulation and it is linked to the constrained least square error (LSE). In UWC, computationally efficient ML estimation algorithm is applied with quadrature phase-shift keying (QPSK) modulated OFDM for start-of-frame (SoF) detection over a typical underwater channel. ML estimation is proposed in time domain. Data interleaving is used to improve the bit-error-rate (BER) performance of the coherent detection scheme at the receiver. Furthermore, simulation results of a BER of $4.8 \times10^{-6}$ for a 512 QPSK symbol preamble length at SNR of 14dB is achieved. It is also shown that mean square error (MSE) is less than $10^{-9}$ for a 512 QPSK symbol preamble length.