Sampled Coding - Dispersion of Reliability

A new approach for concatenation of codes is proposed. Only a part of data symbols is fed to an outer encoder whereas the remaining part is directly forwarded to an inner encoder. (For example the inner code can be intersymbol interference causing process .) Thus, only samples of data are encoded by the concatenated structure. During an iterative decoding procedure is improved reliability not only of the twice coded data, but also of the once coded data. An estimation of the performance of sampled coding scheme is derived and compared with simulation results. Several examples have been found where all data symbols achieve higher reliability in a sampled coding approach than in a conventional concatenated scheme with equal (average) rate. Additionally , there are two (or more) classes of symbol reliabilities. The rst one shows extreme low error rate for the concatenated encoded symbols. The second one shows error rates that are signiicantly lower compared to a scheme employing only the inner code. This eeect can very eeciently be exploited by an appropriate combining of source and channel coding. 1. INTRODUCTION Ensuring high reliability of transmitted data is one of the most important tasks of digital transmission systems. A very important tool to reach this goal is channel coding based on adding redundancy. There are a lot of important parameters of channel codes which characterize channels codes. Two of the most important are code rate R c i.e. (for binary codes) the ratio of informations bits to the total number of transmitted bits complexity of the encoding and decoding procedure. Codes should be used with redundancy as little as possible to guarantee necessary reliability of data. The more redundancy, the greater transmission speed and the more signal bandwidth are necessary, but also the better protection of data is achieved. A popular method to achieve long powerful codes at a moderate complexity is concatenation of codes. Concatenation can be done in a serial and a parallel way (e.g. Turbo-codes). Here, we focus on serial concatenation. Each of the component codes of a concatenated scheme can be very simple, but together they provide a powerful protection of the data. At the receiver side the codes are decoded successively, hence the complexity grows only linear with the number of codes used. Interlea-ving between the encoders usually provide additional performance gain. Up to now, the most powerful coding schemes known are concatenated codes, which …