A Experimental System Used For The Simulation And Improvement Of The Performance Of Turbo Codes

Error correction codes are a means of including redundancy in a stream of information bits to allow the detection and correction of symbol errors during signal's transmission through the channel. The emerge of error correction coding showed that Shannon's channel capacity could be achieved when transmitting information through a noisy channel. Turbo codes, with a PCCC coding architecture and a SISO & iteration decoding strategy, are a very powerful error correction codes that bring the performance of practical coding even closer to Shannon's theoretical threshold. It has been regarded as the milestone on channel coding theory.Discovered recently by computer simulation, turbo codes lack theoretical background. Now theoretic analysis and computer simulation are two primary means of research. It's very difficult to theoretically investigate turbo codes because of its complicated coding technique and decoding strategy. However, Simulation can not only evaluate the validity of turbo codes' theory, but also make guidance on turbo codes design and implementation.In this thesis, turbo codes' fundamental theory is addressed, some techniques used in turbo codes are also given in detail. A MATLAB based turbo codes' software simulation system is constructed, experimental simulations are then comprehensively performed, upon which simulation results are also compared with those published literatures. The bit-error-rate ( BER ) performance of various scheme of Turbo codes are compared also by the use of our softwar simulation system. A further approximation to reduce decoding latency on Log-MAP algorithm is put forward and analyzed, a non-conventional decoding structure is also proposed aiming at reducing the complexity of Turbo codes implementation as well as decreasing time delay of the decoding.