Multiple Symbols Soft-Decision Metrics For Coded Frequency-Shift Keying Signals In Symmetric Alpha-Stable Noise

Communication links are often disturbed by non-Gaussian noise, which has significant impulsive characteristics and a thick tail in its probability density distribution. Impulsive noise has serious negative effects on receivers and signal detection in communication system. The original algorithms and equipments designed based on the Gaussian noise can’t meet the requirements when they deal with the signals interfered by impulsive noise. But we can solve this problem by designing new algorithm based on the statistical physical model of impulsive interference. The current statistical physical model of impulsive noise is symmetric alpha-stable distribution. To adapt the most current wireless communication system adopting Turbo code, it is necessary to design the soft-decision metrics according impulsive interference and enhance the communication reliability. This paper will focus on the soft-decision metrics for coded MFSK signals in symmetric alpha-stable noise.The thesis analyzes the probability density of SaS noise and compares the performance of decision metrics based on Gaussian noise with decision metrics based on impulsive noise by simulation in diversity impulsive interferences. And it designs a novel soft-decision metrics for communication system adopting Turbo. Based on the analysis and comparation of five soft-decision metrics for Turbo coded orthogonal signal in SaS noise, namely Bayesian Gaussian decision, Bayesian Cauchy decision, GLR Gaussian decision, GLR Cauchy decision and GLR Gaussian decision without channel state information, the blind detection metrics for coded MFSK signals in SaS noise is presented and the single symbol soft-decision algorithm is extended to multiple symbols soft-decision metrics.The relationship between the symbol length of multiple symbols soft-decision metrics and the performance and the complexity is analyzed according the performance of multiple symbols soft-decision metrics with diversity symbol length for Turbo coded MFSK signals in different SaS noise. The symbol length is proposed for lower complexity and better performance. After the analysis of the fading features of communication channel and combining the fading characteristics with the multiple symbols soft-decision metrics, the multiple symbols soft-decision metrics with different symbol length for slow block fading channel are studied and the symbol length for excellent performance and low complexity’ is found out. Then it improves the multiple symbol soft-decision metrics and adapts it to fast fading channel with the optimal symbol length. The simulation results of improved multiple symbols soft-decision metrics for fast fading channel show good detection performance.The improved multiple symbols soft-decision metrics combining the channel fading characteristics provides a new way for blind detection in high speed mobile wireless communication system.