Two major issues associated with fiber-wireless technology are the nonlinear distortion of the optical link and the multipath dispersion of the wireless channel. In order to limit the effects of these distortions, estimation, and subsequently equalization of the concatenated fiber-wireless channel needs to be done. This thesis addresses three scenarios in this regard, they are: uplink estimation using pseudonoise (PN) sequences, downlink estimation using Walsh codes, and uplink equalization using a decision feedback equalizer (DFE) and series reversion, all in the presence of both wireless and optical channel noise. The training sequences used in the identification are practically feasible. These training sequences have white noise-like properties which effectively decouples the identification of the linear and nonlinear channels. Correlation analysis is then applied to identify both systems. Furthermore, we propose an algorithm to mitigate the adverse effect of multiple access interference (MAI). Numerical evaluations show a good estimation of both the linear and nonlinear systems with 10 users for the uplink and 54 users for the downlink, both with a signal-to-noise ratio (SNR) of 25 dB. Chip error rate (CER) simulations show that the proposed MAI mitigation algorithm leaves only small residual MAI.