Exploration of Universal Constant-Envelope Predistortion Using Correlation-Based Signal Attenuation, 10-9218Printer Friendly Version
Inclusive Dates: 10/02/00 - 04/01/02
Background - A major problem with multicarrier communication systems is the high peak to average power ratio that results from numerous carrier frequencies going in and out of phase. Because of the nonlinear behavior of electrical components, signals with a high degree of dynamic range (high peak to average power ratio) become distorted when they are transmitted or received. Two main sources of this nonlinearity are the transmitter power amplifier and the frequency translation (or mixing) stages. Because wireless modems, European digital television broadcasts, and cellular base-stations all transmit multicarrier signals, systems using each of these technologies could be improved using a form of digital predistortion.
Approach - The purpose of this study is to explore a method of ruggedizing a signal against this nonlinearity. By predistorting the signal, so that signal energy is transferred from the signal peaks to the signal valleys, the peak power can be reduced without severely limiting the average power. By selectively smoothing (not just clipping) the peaks, the signal may be received with better clarity [lower Bit Error Rate (BER)], because the signal energy is not spread into out-of-channel frequencies by such severe nonlinearity.
There are three main tasks for this project. The first is to create an algorithm for this particular method of digital predistortion for multicarrier communication systems. This method uses the correlation between each signal and the sum of all signals to create a particular time-varying attenuation level for each individual signal. The second task is to optimize the parameters of the algorithm with computer simulation and find the maximum system bit error rate improvement possible through the algorithm for three test scenarios (10 carriers, 20 carriers, and 50 carriers). The third task is to verify the system bit error rate improvement found through computer simulation with a hardware demonstration.
Accomplishments - There are two major accomplishments. The first is the completion of the algorithm, which determines the attenuation for each signal at each point in time. The second is the optimization of the algorithm via computer simulation. Optimal values for all four algorithm parameters (maximum gain, minimum correlation for minimal attenuation, sensitivity of attenuation to correlation, and window size) have been empirically determined for the three test scenarios. Finally, the results were verified by laboratory measurements. An arbitrary waveform generator was used to create the signals, which were upconverted and passed through a radio frequency (RF) power amplifier. The output of the amplifier was attenuated, downconverted, and stored by a digital oscilloscope. The received RF signal was then demodulated by the computer to determine the effects of the algorithm on an unmodified receiver's ability to demodulate the signal.