Underwater acoustic digital communication was investigated by simulation under conditions of various distortions which seem to degrade communication performance, such as multipath or Doppler shift. M-sequence data was used as transmission data. The signal band was 20 kHz±4 kHz, and the modulation method was 16-quadrature amplitude modulation (16-QAM). The received data was demodulated using a multichannel decision feedback equalizer (DFE). In this paper, the effect of feedforward tap allocation on communication performance under an intensive multipath condition is considered. It became clear that the communication performance is advanced by utilizing multipaths, that this advantage is distinguished when the signal-to-noise ratio (SNR) of the input signal is small, that it is necessary to consider the trade-off between this advantage and the instability caused by a large number of taps, and that this advantage is available if ever there is Doppler shift or few multipaths.

Multiuser communication is demonstrated using experimental data (450-550 Hz) collected in deep water, south of Japan. The multiple users are spatially distributed either in depth or range while a 114-m long, 20-element vertical array (i.e., base station) is deployed to around the sound channel axis (~1000 m). First, signals received separately from ranges of 150 km and 180 km at various depths are combined asynchronously to generate multiuser communication sequences for subsequent processing, achieving an aggregate data rate of 300 bits/s for up to three users. Adaptive time reversal is employed to separate collided packets at the base station, followed by a single channel decision feedback equalizer. Then it is demonstrated that two users separated by 3 km in range at ~1000 m depth can transmit information simultaneously to the base station at ~500 km range with an aggregate data rate of 200 bits/s.

Figure 2. Normalized metallicity distribution functions (MDFs) derived from observations (histograms) and predicted by our model. Upper panel: the dashed, gray, and solid black lines show the results directly taken from the output of our simulation, using a Gaussian dispersion in [Fe/H] with width σ = 0 (no dispersion), 0.05, and 0.17, respectively. The histograms represent different sets of observations-red: Casagrande et al. (2011); blue: Adibekyan et al. (2012; their MDF can be found in Rix & Bovy 2013); orange: Bovy et al. (2012). Lower panel: the orange, black, and gray lines show our predictions after applying the correction for low-mass stars (see the text). Each histogram represents a combination of two sets of observations.

An experiment on high-speed underwater acoustic data communication was carried out at real sea as a time-variant multipath channel. Demodulation was processed by a multichannel decision feedback equalizer (DFE) demodulator with adaptive filtering. The modulation method used was 16-quadrature amplitude modulation (16-QAM), transmitting rate was 32 k bit per second (bps), carrier frequency was 20 kHz, and bandwidth was 8 kHz (16–24 kHz). Good demodulation results were obtained at a sea trial despite the presence of high-level multipaths. At real sea, we proved that high-speed (32 kbps) acoustic communication can be realized even in a multipath fading channel with 16-QAM. Therefore, it is expected that the flexibility of the design of underwater equipment will be improved, because the influence of a circumferential structure can be neglected.

An at-sea experiment of multiuser communication in deep water using adaptive time reversal was carried out. In the experiment, two sources and a 20-channel receiver array were deployed at the range of 30 km in an area of 1500-2000 m water depth. One of the sources was moored and the other was suspended at various depths. For processing signals actually transmitted from two sources, it was demonstrated that adaptive time reversal could cancel multiuser interference independently of relative source positions. Additionally, for a more detailed investigation, the analysis of multiuser test signals by synthesizing signals from different depths was performed. As a result, especially in the case of adjacent sources, adaptive time reversal significantly suppressed crosstalk.