Free space optical communication systems with multiuser diversity. (c2017)

Abstract

We explore the performance of N-user multi-point-to-multi-point Free Space Optical (FSO) communication systems. For such systems, the communication between the N transmitters and N receivers relies on user grouping for the sake of achieving enhanced spatial diversity levels. Unlike non-cooperative systems where each one of the N users sends its own information along the corresponding FSO link independently from the information sent from the other users, joint encoding/decoding is applied in the multiuser cooperation scheme. In this case, the N information symbols are jointly encoded and transmitted along a selected number of FSO links comprising the N-th, (N-1)-th... best channels. We provide a bit error rate (BER) analysis of the above system over gamma- gamma atmospheric turbulence channels. The unipolar (Non negative) M- ary Pulse Amplitude Modulation (PAM) with intensity modulation and direct detection is the assumed modulation technique in the targeted system. We approach the grouping scheme as an optimization problem to derive the best bit allocation strategy. The recurrence relations that exist between the Probability Density Functions (PDFs) of order statistics is used to introduce approximate expressions of those PDFs which results in accurate expressions of the Bit Error Rate (BER). The user grouping vector strategy can be described by the N-dimensional vector G whose n-th component represents the number of M-PAM information symbols that are allocated to the n-th best channel.The results show that FSO systems performance is highly dependent on the grouping strategy. For a system with N = 2 users, the grouping G = [0 2] outperforms the non-cooperative scheme performance (G = [1 1]). In other words, the best strategy corresponds to allocating two 2-PAM symbols to the best channel. For a system with N = 3 users, the grouping G = [0 1 2] outperforms the non-cooperative scheme performance over all SNR ranges. However,it shows better performance than the All-user grouping G = [0 0 3] only below 36 dB. For N = 4 users the best performance is recorded for the grouping G = [0 0 1 3]. For N = 5 users the grouping G = [0 0 1 2 2] shows the best performance for SNR below 26 dB while the grouping G = [0 0 0 2 3] results in the best BER for SNR higher than 26 dB. In Chapter 1, we introduce the FSO technology by stating its advantages, limitations and applications. We also introduce in this chapter the modula- tion technique used and the Gamma-Gamma model. We show the diversity methods used in the literature then we describe in Chapter 2 the multi-point- to-multi-point proposed scheme. In this Chapter, we evaluate the di erent BER expressions for the nth best user using order statistics. Chapter 3 shows the results and simulations obtained which validates the theoretical analysis. We conclude the best grouping scheme for each system and classify users into categories.