ROBUST TIMING SYNCHRONIZATION FOR MOBILE OFDM AND COGNITIVE RADIO BASED SYSTEMS

• Main Author: Suyoto
• Format: Dissertations
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/39981
• Institution: Institut Teknologi Bandung

High data transmission in mobile communications faces several challenges, due to high delay spread and rapid change of wireless channel. Orthogonal Frequency Division Multiplexing (OFDM) is a suitable modulation technique for applications with high delay spreads. However, OFDM performance is very sensitive to time synchronization errors. The performance of OFDM systems is influenced by timing synchronization, channel estimation, and mobility. Errors in time estimates not only cause InterSymbol Interference (ISI) but also degrade the performance of OFDM system as a whole. Requirements for synchronization will be high with increasing complexity system. For example, in a moving environment the effect of delay spread will increase and channel links change rapidly. The autocorrelation method using training symbol is the most popular algorithm for synchronization because of its low complexity so it’s easy implemented. However, the autocorrelation method has weaknesses in terms of robustness in dealing with multipath channels. Improvement methods, although improving performance, are generally not robust to multipath channels with high delay spread. While time synchronization methods that do not use training symbol (blind time synchronization) are very complex and low performance in multipath channel. Related to these weaknesses, the focus of the research carried out was on problems related to robust time synchronization algorithms in dealing with multipath channels with a high delay spread and done by using training symbol. Time synchronization is done in 2 stages. In stage 1 (time synchronization coarse) is done by calculating autocorrelation from the symmetrical correlator, then the results of the autocorrelation signal are mathematically modelled as a binary hypothesis testing problem between conditions there is no training symbol (H0) with conditions there is a training symbol (H1). The proposed method is carried out solving the binary hypothesis testing problem with the difference in the PDF (Probability Density Function) function of the symmetrical correlator signal received. Mathematically derived that the PDF is different in both conditions. Furthermore, the algorithm is proposed to distinguish it by using normalization of high order central moment. Distribution when there is a training symbol has a higher mean of distribution when there is no training symbol. The proposed coarse time synchronization method calculate the central moment and iii compares it with its average value. After getting the maximum value then compared to the threshold. In stage 2 fine time synchronization is done by using the cyclic shift technique from channel estimation, so that overall synchronization time performance is better. Algorithm testing is done by evaluating its performance with the size of MAE (Mean Absolute Error) and MSE (Mean Squared Error) in multipath channel environments with high delay spread. The evaluation results performed on multipath channels with high delay spread (vehicular B channel) obtained a gain greater than 25 dB in the MAE test and obtained a gain greater than 20 dB in the MSE test. Furthermore, we also tested the algorithm on OFDM-based Cognitive Radio (CR) systems, where the system was interrupted by narrowband interference (NBI) and wideband interference which was high. Evaluation results conducted on vehicular B channel with a high NBI (SIR= 0 dB) obtained a gain greater than 20 dB from the autocorrelation method in the MAE test. Evaluation results conducted on vehicular B channel with a high wideband interference (SIR= 0 dB) obtained a gain greater than 20 dB from the autocorrelation method in the MAE test. In addition to having good performance in multipath channel with high delay spread, the proposed algorithm is also good against NBI, wideband interference, and can adapt to the use of fewer pilots.