Timing acquisition of frequency-hopped ultra-wideband OFDM system

In this thesis, the timing acquisition of a frequency-hopped (FH) ultra-wideband (UWB) orthogonal frequency division multiplexing (OFDM) system is studied. A scenario in which the time-frequency code (TFC) is deterministic but unknown to the receiver is addressed. Acquisition time and the training data overhead should be kept to a minimum. This makes timing acquisition in the FH system a challenging task. First, the FH-OFDM technique is presented and the stages in timing synchronization of a general wireless communication system are introduced. Existing methods for decision variable generation, detection of the presence of the signal, and determination of the TFC are reviewed. Signal design for synchronization and timing synchronization requirements are discussed. After presentation of the target system model’s characteristics, simulation goals and assumptions are specified and simulation results are provided and analyzed. First, simulations are run for an additive white Gaussian noise (AWGN) channel, and then the effects of a frequency-selective fading channel are studied. Data-aided methods for decision variable generation are addressed. Differential correlation is chosen as the basis for the timing acquisition scheme, due to its good tolerance for multipath fading and lower complexity. The probability of detecting the presence of the signal is simulated for different training symbol structures and parameters. A method for determining the TFC is developed, and its performance is simulated for different numbers of subbands and training symbols. The method uses an exclusionary strategy in which the initial list of possible TFCs is shortened iteratively on the basis of results from testing of different subbands for the presence of the signal. The simulation results indicate the detector parameters and preamble structure that are feasible for the target system. The results imply that the differential-correlation-based detector performs well in both AWGN and fading channels and that it is very reliable even at low signal-to-noise ratios (SNR), of 0–5 dB. The TFC determination method developed is demonstrated to be able to determine the correct TFC efficiently even from a set of 30–60 possible TFCs by using a relatively small number of training symbols. As a conclusion, it is suggested that a differential-correlation-based detector with correlation length equal to correlation distance and a search step size of ¼ or ½ symbols should be used. The preamble should contain a number of training symbols of at least twice the length of the TFC.