Study on Improving the Performance for Block Turbo Codes by Dynamically Estimating Weighting Factors Based on Mutual Information

• Main Authors: CHI,CHIA-HUNG, 紀佳宏
• Other Authors: LIN,YI-NAN
• Format: Others
• Language: zh-TW
• Published: 2014
• Online Access: http://ndltd.ncl.edu.tw/handle/d5775k

碩士 === 明志科技大學 === 電子工程系碩士班 === 102 === In a general decoding algorithm of block turbo codes(BTCs), the weighting factor α and reliability factor β are fixed in each static decoding iteration with increasing times. It’s difficult to decode codes under poor channel environment. However, the time-variant channels predefined α and β are not appropriate in BTC decoding because they are not easily being controlled. The thesis proposes a method to set the values of α and β dynamically based on input and output values of the decoder. BTC decoding algorithms use an iterative decoding procedure to achieve low bit error rate performance typically with eight half decoding iterations. However, the requirement for the number of decoding iterations is related to the channel; i.e. , the larger the noise is, the more the number of iterations is required. So in the traditional BTC decoders, we practically measure the input and output values of decoders, then establish entropy and mutual information concepts to make the decoding process work smoothly.