13-06-2012, 04:34 PM
Matlab Toolbox For Turbo Code Simulation
Matlab Toolbox For Turbo Code Simulation.pdf (Size: 387.2 KB / Downloads: 324)
Abstract
Since C.E. Shannon’s publication of his work on transmission efficiency in 1948 [1]
much time and effort has been devoted to the development of error correction codes capable
of allowing error-free transmission of data at a minimum Signal to noise ratio. With our
societies increasing reliance on computers and digital communications this search for the
most efficient transmission codes possible has proved more vital than ever.
Turbo Coding and Telecommunications
Overview
As computers become more and more involved in our everyday lives
communications moves more and more from its roots in analog transmission into a purely
digital form and the key to cheap and effective digital communications is fast, error-free
transmission of data through the established telecommunications networks. Unfortunately
due to the non-ideal nature of the current networks, noise is always a problem, leading to
errors in received data and delays whilst erroneous data is resent.
A brief history of Channel Coding
Following C.E. Shannon’s publication of his work on transmission efficiency in
1948 [1], It became apparent that another method must exist if the values calculated using
Shannon’s limit theorem were true. This led to research into error correction methods, which
in turn led to the creation of Forward error correction (FEC) [2] codes. These work upon the
theory that even if noise alters or destroys the transmitted data, if the receiver can, in some
way, recover the original data from the received data then no problems will occur, in effect
trading off signal power against hardware requirements and decoding complexity.
Enter Turbo-Coding
In 1993 Berrou, Glavieux and Thitimajshima [3] developed a more data efficient
way to combat the effects of burst error in convolutional code. By using two convolutional
coders, one working on a randomly sorted or interleaved version of the same data it was
hoped to lower the chance of burst error destroying all the data related to a set of code words,
given Gaussian random noise it would be more than likely that enough data would exist to
recover the original transmitted data. This method of coding would require a much more
complicated method of decoding however, and a recursive decoding method was developed
utilizing two MAP decoders in series passing data back and forth between them. The
experiment was a success and Turbo Coding has proved to be the most efficient coding
method yet developed, capable of approaching closer to the Shannon limit than any other
coding method whilst keeping overhead to a minimum.