Research On The Peak To Average Power Ratio Reduction For OFDM System By BFGS

OFDM(Orthogonal frequency Division Multiplexing) is an very attractive technique for wireless high-bit-rate transmission due to the minimizing effects of frequency-selective fading and its high bandwidth efficiency. Howerer, OFDM has the high PAPR(Peak-to-Average-Power Ratio) which is its inherent disadvantage. These large peaks cause saturation in high-powered amplifier,leading to the lower transmitted efficiency, and increasing the complexity of A/D and D/A convertor.Therefore,the high PAPR limits the practical applications of OFDM.So,the PAPR reduction is of viral importance for OFDM systems with power efficient transmitter.Now,there are some technologys to solve it,such as signal pre-distortion techniques, and probability techniques. And two typical methods of probability techniques are SLM(Selective Mapping) and PTS(Partial Transmit Sequence) , which is widely used.Although SLM and PTS has little distortion,the computation complexity is quite large for the extra IFFT calculations. Then, IPM(interior point methods)has been propo- sed.IPM ,also refered to as the barrier methods, are a class of algorithms to solve linear and non-linear convex optimization problems, and through several Iterative convex programming Algorithm, optimal solution could be achieved faster than PTS or SLM.In this thesis, firstly,I would do research on the IPM algorithm. And based on Constellation EVM(Error Vector Magnitude) and power overhead, barrier function would be built.Secondly,I would do research on the recent algorithms of lowering the PAPR of OFDM signals by IPM , and then present BFGS algorithm to solve the IPM and get the global optimization.Different from the general IPM algorithms ,which use Newton methods to achive the optimal value, BFGS builds up an approximation of the inverse Hessian,so the complexity is reduced. Through MATLAB simulation , the results verify BFGS algorithm reduces PAPR value effectively, but the complexity is lower then the traditional Newton methods.