Traditional OFDM implementations use conventional Fourier filters for data modulation, via the IFFT operation. Much research suggests wavelet based OFDM provides performance gains, due to superior spectral containment properties of wavelet filters. When compared in simulation, neither system was found fundamentally better than the other. Rather than focus on filter bank design, joint detection strategies provide a relatively simple method for improving system performance. The conventional matched filter used in OFDM systems, is only optimal when the channel introduces no carrier distoration. We propose an optimal detector, which maximizes the probability of making correct decisions, and a suboptimal method which decorrelates carriers based on knowledge of the channel. Both joint detection methods provide significant performance gains in OFDM systems over conventional matched filtering.