Wavelet Modulation (WM)---also referred to as fractal modulation---simultaneously sends data at multiple rates through an unknown channel. This novel multirate diversity strategy offers improved message recovery over conventional modulation techniques: if the message is not received at one rate due to the channel disturbances, it can be received at another rate where the channel is clear. Previous research has demonstrated the performance of wavelet modulation in Gaussian channels.

This paper extends the investigation to the performance of wavelet modulation in time varying channels. We show that bit error rate (BER) wavelet demodulation performance in the additive, white, Gaussian noise (AWGN) channel is comparable to theoretical binary phase shift keying. Our results depict the improvement in performance that can be realized for time varying channels by utilizing the various rates of wavelet modulation; BER versus SNR curves for each fading channel show how the message at one rate can be more accurately recovered than at another rate. Furthermore, we also present results indicating the dramatic improvement in wavelet demodulation performance when multiple rates are available and utilized for demodulation. This improvement in performance is highly visible in BER performance for the AWGN and flat fading channels.

Finally, a comparison of binary phase shift keying (BPSK) and WM in a frequency selective channel is performed. The BER improvement of WM is shown when demodulation is done at rates which are least corrupted by ISI. We illustrate our new algorithm that: identifies the channel characteristics; determines which rates are maximally corrupted by ISI; and, utilizes only those uncorrupted copies in demodulation.