The performance of two concatenated coding systems using a K=3, R=1/2 convolutional inner code and a Reed-Solomon (RS) (15, 9) or (15, 7) outer code was measured over a 5.76-km-long atmospheric direct detection optical communication channel. Inner code interleaving of 100 mu s combined with outer code interleaving of 240 bits (60 RS symbols) was found to be sufficient to obtain a decoded BER of less than 10/sup -6/ under conditions of moderate channel turbulence and an average of 6-10 detected photons per channel bit.>

The performance characteristics of a direct-detection optical communication system, in which a laser transmitter sends short optical pulses of selected nonoverlapping center frequencies, are analyzed. It is shown that this signaling scheme can be used in conjunction with ReedSolomon coding to obtain high data rate, reliable communication under conditions of optimal energy efficiency.

The performance of both optimal and much simpler suboptimal direct detection optical communication receiver structures was directly measured and found to be nearly identical for laser intensity subcarrier PSK communication through both free space and atmospheric (log-normal fading) channels.

Experimental results of estimating the delay of optical pulses in the presence of background noise based on the occurrence times of photoelectrons emitted by a direct detection photoelectric effect receiver are presented. Both minimum mean square error (MMSE) and maximum likelihood estimation techniques were used. The MMSE estimates of the delay of both Gaussian and rectangular optical pulses were obtained from the evolution of the posterior conditional probability density of the pulse delay given the photocount occurrence times. For signal to noise intensity ratios as low as 10 dB, it was found that the experimental performance of the MMSE delay estimates remained near performance lower bounds and that a rectangular pulse gave better performance than a Gaussian pulse of equal signal energy and equal peak signal intensity. At lower signal to noise ratios, the performance of the estimates for both pulse shapes deteriorated rapidly. It is shown that, in the absence of background noise, MMSE and ML estimation are equivalent.

The effects of photon counter dead time on a direct detection binary pulse position modulation optical communication system are computed for both a free space optical channel and an atmospheric channel subject to clear air turbulence. The degradation in the performance of an optimum receiver structure due to dead time losses was found to be typically 1 dB for the free space channel and less than 1 dB for the atmospheric fading channel, even at the large value of the ratio of counter dead time to counting interval length of ten percent.

The experimentally measured performance of equations which govern the evolution of both causal minimum mean-square error (MMSE) and approximate (suboptimal) MMSE estimates of the intensity function of a doubly stochastic Poisson process for the case of log normally faded optical field intensities is reported.