In the uplink of a fiber-based wireless system, the multipath dispersion that is introduced by the wireless link and the nonlinear distortion that is caused by the radio-over-fiber (RoF) link significantly degrade system performance. Channel equalization is challenging, because these impairments are generally unknown to the receiver. One novel technique of estimating both of these impairments using ternary pseudorandom signals is proposed. The key idea is to exploit proper harmonic suppression through signal design such that errors that are inflicted by the nonlinear distortion can be eliminated or reduced. Capitalizing on this case, a simple yet effective correlation approach is applied to accurately estimate the channel impulse response. Subsequently, least squares polynomial fitting is invoked to identify the nonlinearity of the RoF link, making use of the linear estimates obtained earlier. Design guidelines that lead to good performance are proposed. Results reveal that an improvement in the estimation accuracy of at least a factor of two is achieved by the proposed technique over an existing technique utilizing binary signals in the various scenarios investigated, even if the identification period of the proposed technique is almost 1000 times shorter than the existing technique. Because the proposed technique enables a simpler transmitter structure and requires shorter identification periods and lower power transmission, it is found to be appealing for use in broadband fiber-based wireless systems.