Thresholding-Based Optimal Detection of Wireless Optical Signals

The receiver design for a high-speed free-space (wireless) optics (FSO) signal is necessarily highly complex when channel state information (CSI) is not available. Currently, although most approaches provide high detection performance in terms of bit error, receiver design is difficult to implement. This paper proposes two practical thresholding-based detection schemes, which offer significant improvement to receiver throughput on a computational load basis when CSI is not available. The first is based on a simple maximum likelihood (ML) function where the bit error rate (BER) is the same as conventional symbol-by-symbol detection. This method, however, causes a loss of BER performance. The second uses the aid of pilot-symbol-assisted modulation (PSAM) to modify the ML function when channel coefficients are temporally correlated. While numerical analysis based on this method shows that the BER performance in a log-normally distributed fading channel is very close to detection achieved with perfect CSI, the receiver suffers from increased complexity. If random processes for fading and noise are assumed as stationary and given that the detection threshold is quickly calculated and applied during a given period, such complexity of PSAM-based and symbol-by-symbol detection methods can be reduced.