Search for OFDM synchronization waveforms with good aperiodic autocorrelations

The problem of searching for OFDM synchronization waveforms with good aperiodic autocorrelation properties is addressed. A synchronization waveform is embedded within data in the form of pilot tones. Selection of the sub-channel locations to be reserved for synchronization pilots is critical, as each different selection results in a waveform with a different sidelobe pattern. The optimal pilot tone location selection problem is mathematically intractable and an exhaustive search for practical systems is infeasible. A genetic algorithm is employed for searching the locations of a number of pilot tones among all OFDM subcarriers, such that the resulting time waveform has good autocorrelation properties. The performance of the generated synchronization waveforms was evaluated in a simulated OFDM system and it was demonstrated that they achieve the desired frame synchronization performance. It was observed that the genetic algorithm accomplishes the task much faster than a random search.