摘要 |
Each input binary digit of an input serial bit string having a referential word in a referential word time period is sampled N times to produce a plurality of over-sampled binary digits corresponding to a first group of sampling operations, a second group of sampling operations,--, an N-th group of sampling operations, and the over-sampled binary digits are divided to N divided bit strings corresponding to the N groups of sampling operations respectively. Because each divided bit string having the referential word is correctly sampled at high probability, a word start position of the referential word in each divided bit string is detected, one divided bit string correctly sampled at the highest probability is selected, and a string of words starting from the word start position is retrieved from the selected divided bit string and is output.
|