Research On The Peak-to-Average Power Ratio Reduction In Broadband Wireless OFDM Systems

Due to high spectral efficiency, good robust to multi-path fading, and high-speed datatransmission, OFDM has always attracted much attention. Now OFDM is regarded as oneof the key techniques for next-generation mobile communication systems. However, as amulti-carrier technique, OFDM signal is the result of superposition of many individualsubearriers, which leads to large PAPR. The signal with large PAPR is affected by thenonlinearity of the high power amplifier at the transmitter, causing performancedegradation. Therefore, PAPR limits the application of OFDM technique to practicalsystems, and how to reduce PAPR has become a research hotspot in communication area.With the support from the National Natural Science Foundation of China (NSFC) underGrant No. 60496315, this thesis investigates the PAPR in OFDM systems.Firstly, this thesis derives a novel companding transform based onβ-distribution toreduce the PAPR of OFDM signals. The key idea is to transform the distribution of thereal (imaginary) part or amplitude of original OFDM signals to a boundedβ-distribution.Moreover, given different parameters, the proposed companding transform has variousexpressions. It is proven that the well-known companding schemes based on error functionand exponential function are actually special cases of the proposed companding transform.Additionally, this thesis derives another simpler expression of the proposed transform. Thesimulation results show that the novel companding transform can significantly reduce thePAPR while maintaining the average power of signals unchanged. Moreover, the novelscheme offers better system performance than the well-knownμ-law, modifiedμ-law,error function, and exponential function companding transforms.Secondly, this thesis proposes two novel modified flipping algorithms. One is tomodify the iterative fashion of the flipping algorithm. The main idea is that the formedPTS sequences are partitioned into different groups and then the iteration is performedgroup by group. The modified algorithm is more flexible for PAPR reduction. Thesimulation results show that different group partition has different PAPR reductionperformance better than the original flipping algorithm. In fact, the original PTS andflipping schemes can be regarded as the special cases of the modified algorithm. In theother modified flipping algorithm, the transform of the PTS sequence is extended. Five simple sequence transform are deliberately chosen. At the receiver, the original input datacan be easily recovered by means of Fourier transform properties, and thus avoidingcomputational complexity increase. The simulation results demonstrate that with similarcomplexity, some of the five transforms offer better PAPR reduction performance thanthe original flipping algorithm, and even than the PTS scheme when the number ofsubblocks is not large.Thirdly, this thesis investigates the unitary transform for PAPR reduction. Someearlier researches showed that performing the Hadamard transform on the input symbolsequence could lead to a lower PAPR. However, they did not give the theoretical analysis.This thesis proves that performing unitary transforms including Hadamard transform onthe input sequence can reduce its aperiodic autocorrelation and as a result, PAPRreduction can be obtained. Moreover, this thesis analyzes BER and PSD performance ofunitary transformed signals, which shows that the unitary transform has no impacts onBER and PSD. The simulation results invalidate our analysis. Compared with clippingand companding methods, unitary transform causes no BER and PSD performancedegradation. Compared with SLM and PTS schemes, unitary transform has much lowercomputational complexity and no side information is needed. Therefore, the unitarytransform scheme is a promising technique for PAPR reduction.Finally, this thesis studies the PAPR in MIMO-OFDM systems, and analyzes theperformance of STBC MIMO-OFDM systems. Two different clipping schemes arecompared and it is found that clipping before STBC can provide better systemperformance. For this clipping scheme, the SER and BER expressions of the clippedsignals with different modulation formations over a multi-path fading channel arederived. Both the simulation results and theoretical analysis show that the clippingoperation can be directly applied to the MIMO-OFDM system for PAPR reduction at thecost of system performance degradation. Additionally, it is shown that the scheme ofclipping before STBC exhibits better system performance and more immunity to clippingratio variation. Therefore, clipping before STBC is more suitable for STBCMIMO-OFDM systems.