Free-space-optical communications over turbulent channels. (c2009)

Abstract

Local Area Networks (LAN’s) are being expanded over large geographic areas and they are used in many business fields. Thus, communication links between buildings are to be optimized in order to achieve high transmission rates, high performance levels, low cost and ease of deployment. Free-Space Optics (FSO) is a communication system that achieves all of the above objectives and, thus, constitutes a strong candidate solution for such networks. FSO systems are based on transmitting information through light beams in free atmosphere and they suffer from fading due to atmospheric scintillations. Fading effects can be reduced by deploying laser arrays and photo-detector arrays at the transmitter and receiver sides respectively. Such systems are referred to as Multiple-Input-Multiple- Output (MIMO) FSO systems. In addition to their robustness against fading, MIMO FSO systems can also enhance the data rate since the array of lasers can be driven by independent information sources. Fading over FSO channels is often modeled by either Log-Normal or Rayleigh distributions. In this context, the first contribution of this work consists of an analytical characterization of the diversity order that can be achieved by MIMO FSO systems over such channels. Based on the Erlang approximation, closed-form expressions of the error-rate and channel-capacity were derived. These simple expressions offer useful insights on the performance gains that can be achieved at a given Signal-to-Noise Ratio (SNR). On the other hand, for estimating the values of the transmitted signals, exiting MIMO FSO systems are often associated with Maximum-Likelihood (ML) decoders. Although these decoders achieve the smallest error rate, they suffer from an increased complexity since the required decoding time increases exponentially with the size of the transmitted constellation. The second contribution of this work consists of proposing two novel simplified ML decoders that reduce the processing time without increasing the error rate. We also propose suboptimal versions of these decoders that present the advantage of very fast convergence times at the expense of a slight increase in the error rate. All the presented analysis and designs are supported by simulations and analytical proofs.