Synchronization Technique And Synchronization Algorithms In OFDM Systems

Orthogonal frequency division multiplexing (OFDM) is a multi-carrier modulation technology, of which the carrier wave orthogonalizes with each other to efficently use the spectrum resources and the total bandwidth is divided into a number of narrowband subcarriers so that frequency selective fading can be effectivly against. However, because of the overlapping of OFDM subcarriers spectrum, the frequency shift can destructs the orthogonality of the subcarriers which causes the interference among them. Therefore, the estimation of synchronization error for such system is particularly important.Chapter 1 introduces the development of mobile communication. Mobile communication origined from the early twentith century, began to develop in 70's and now has achieved big success. It can be divided into three stages, from the first generation of analog communication systems (1G) to the second generation of digital communications systems (2G), and to the third generation mobile communication systems (3G) at present. Then the status quo and techincal characteristics of OFDM technology is introduced. OFDM techonlogy has wide application due to its numerous irreplaceable advantages. Finally, the existing synchronization algorithms are briefly described.Chapter 2 studied the basic principles of the OFDM system, analyzed the theoretical basis of the orthogonal between carriers. Cyclic prefix can eliminate the interference among carriers caused by multi-path delay so that the orthogonality among them is maintained. Discrete (fast) Fourier transform and inverse transform applied in OFDM has greatly reduced the complexity of system. Then several key techniques of the OFDM system have been described, such as PAR, channel coding and interleaving, channel estimation and synchronization in OFDM systems applications.Chapter 3 is the analysis of synchronization technique in OFDM system. Firstly, the principle and implementation process of synchronization for the OFDM system is analyzed briefly. Secondly, the serious influence on the system caused by symbol timing error, carrier frequency offset and sample deviation is introduced. symbol timing is the most critical factor for an OFDM system, its accuracy will affect the estimation of frequency deviation. Add cyclic prefix in the front of the sent symbols and set its length so that only the added cyclic prefix part of the symbols can be interfered by multi-path delay and the data signals will not be interfered, which can maintain the orthogonality among subcarriers. Carrier frequency synchronization is the most sensitive part of the system, so it should maintain the synchronization among the carrier frequecy. The influence of carrier frequency offset can be divided into two aspects:(1) Integer multiple frequency offset produces a rotating phase of the received signal, which must be compensated, otherwise it would cause the bit error rate increase to a half. (2) Fraction multiple frequency offset in coherent demodulation, the value of other subcarriers at the sampling point is not zero, which will interfere with the demodulated signal, actually, that is, orthogonality among the carriers will be destructed.Chapter 4 gives the deeply research and analysis into the synchronization algorithm based on training sequence. Schmidl & Cox algorithm use two training symbols to complete symbol timing and frequency offset estimation. The first training symbol, whose front part and back part is same, use its good correlation property to search for the location of the maximal relevant sum as the timing point. Since the presence of cyclic prefix, measure function has a platform with the same length of the cyclic prefix, which cause a greater influence on the synchronization precision. Frequency offset estimation uses the first training symbol carrying out fraction multiple frequecy offset estimation according to the timing estimation, and then carry out integer multiple frequency offset estimation based on the differential relationship between the second training symbol and the first training symbol. The principle of Minn algorithm is to add negative values to the related items of mesure function so that eliminate the so-called platform area described above. The peak time of the measurement function improves the estimation accuracy, but the time points near the peak value and close to the maximum peak value affect the timing estimation performance. Park algorithm uses the self-relation of training symbols to complete the synchronization. Designment of completely different sum structure for measure function relative to Schmidl & Cox algorithm. During calculating the sum items of measure function, the sum items of a function before and after are different so that the function value will mutate at the location of right time point resulting in a sharp peak, thus the timing accuracy is improved greatly. However, due to the designing structure of this training symbol, the measure function will have two side peaks which affect the timing accuracy.Chapter 5 studies of an improved synchronization algorithm based on training sequence. Park synchronization algorithm designs a new training symbolsis based on Schmidi & cox algorithm, the proposed symbol synchronization method eliminates the described platform area, but the measurement function exists side peak about N / 4 point of the time point, which affects the timing synchronization estimation accuracy. In order to eliminate this side peak, an improved synchronization algorithm is proposed. Design a new training series to eliminate Schmidi & cox algorithm platform area while not produce side peaks based on the angular designment of sum items related to the measure function. Improved symbol synchronization algorithm use the two sum parts with the same phase corresponding to the front and back parts of the training symbols to work out the relevant sum, which make the measure function produces peaks at the right time point and generates zero values in other points where the sum items with random phase results in the sum value of zero. The pinciple of Frequency offset estimation is consistent with that of Park synchronization algorithm, the fractional frequency offset estimation also use the correlation between the front and back part of the training sequences to work out the fraction frequency offset after the completion of timing synchronization. The integer frequency offset estimation gain the estimation valules by working out the maximal relevant sum values between the local PN sequence and restored FFT demodulation frequency domain training sequence in the receiver. Finally, comparation of the performance between the improved algorithm and several other synchronization algorithms is carried out, and the results show that the symbol timing and frequency offset estimation performance of the improved algorithm are improved relative to other algorithms. Therefore, the improved algorithm is feasible.Chapter 6 summarizes the work done in the thesis and brings forward the future prospects.