We derive a lower bound on the capacity of discrete-time Rician-fading channels that are selective both in time and frequency. The noncoherent setting is considered, where neither the transmitter nor the receiver knows a priori the actual channel realization. Single-input single-output communications subject to both average and peak power constraints are investigated. The lower bound assumes independent and identically distributed input data and is expressed as a difference between two terms. The first term is the information rate of the coherent channel with a weighted signal-to-noise ratio that results from the peak-power limitation. The second term is a penalty term, explicit in the Doppler spectrum of the channel, that captures the effect of the channel uncertainty induced by the noncoherent setting. The impact of channel selectivity and power constraints are discussed, and numerical applications on an experimental Rician channel surveyed in an underwater acoustic environment are also provided.

In order to provide a concise time-varying SISO channel model, the principle of maximum entropy is applied to scattering function derivation. The resulting model is driven by few parameters that are expressed as moments such as the channel average power or the Doppler spread. Physical interpretations of the model outputs are discussed. In particular, it is shown that common Doppler spectra such as the flat or the Jakes spectrum fit well into the maximum entropy framework. The Matlab code corresponding to the proposed model is available at http://perso.telecom-bretagne.eu/fxsocheleau/software.

In passive sonar, bearing tracks that are used for bearings-only tracking (BOT) are sometimes contaminated by erroneous measurements or outliers which may severely degrade performance of BOT when standard unweighted least squares are used. Robustness versus outliers can be significantly improved by Huber's (1964) minimax robust approach or by an efficient discarding of outliers based upon a Viterbi dynamic programming algorithm. Results of computer simulation are shown and discussed.>

Estimating position and velocity of a moving source requires at least an ownship maneuver when performing bearing-only tracking (BOT). By concurrently processing bearing and Dopplerized radiated frequencies a unique target motion analysis (TMA) solution is available without any maneuver. A simple, fast, non-time-recursive estimator, working on several unknown measured spectral lines, is presented. The statistical efficiency of such a Doppler and bearing tracking (DBT) is proved by computer simulation. Simulation results show that DBT can yield proper estimates of source trajectory: range accuracy better than 10% within at most 10 min for a source at 20 km and nine sinusoids.>