Research On Applications Of Turbo Principle To Channel Modulation And Equalization

Turbo codes proposed by C.Berrou el. at. in ICC'93 are a class of powerful channel codes. Their iterative decoder works like a turbo engine so that it is named after turbo codes. Turbo code attracts the significant coverage of researchers in information theory, since turbo codes can achieve remarkable error performance at low signal-to-noise rate close to the Shannon capacity limit. Turbo codes consist of two component convolutional codes concatenated by an interleaver. The difference between turbo codes and traditional channel codes is that turbo codes almost realize random coding by the interleaver and increase the effective block length greatly. Furthermore turbo codes use iterative decoding strategy, which exchanges the information between two component decoders in iterations and approach asymptotically the performance of max likelihood decoding. In academic fields, turbo codes not only are a class of performed-well error correct codes but also provide a way to theory limits. So the principle of turbo codes and its application become a research hotspot. This thesis focuses on the applications of turbo principle on modulation and equalization.Firstly, the algebra and graphical presentation of convolutional code, distance properties and its trellis-based decoder are introduced. On this basis, we analyze the encoder, iterative decoder and performance of turbo codes and serial concatenated convolutional codes. The turbo principle is generalized: the encoder consists of component codes concatenated by interleavers and the iterative decoder is characterized by information exchange.Secondly, the application of turbo principle on modulation is researched. Turbo codes are a class of good error correction codes, however the powerful binary coding schemes are not suitable for bandwidth limited communication systems. In order to achieve simultaneously large coding gains and high bandwidth efficiency, a general method is to combine turbo codes with high bandwidth efficient modulations including trellis coded modulation. We introduce trellis coded modulation technology and generalize four schemes combining turbo codes with high bandwidth efficient modulation: Pragmatic binary turbo coded modulation, multilevel turbo coding, turbo trellis coded modulation (TTCM) with puncturing of parity digits and TTCM with puncturing of systematic bits. On the basis of TTCM with puncturing of parity digits, we propose a novel TTCM scheme based on simple time-varying trellises. Its encoder consists of several two-state time-varying trellis codes concatenated by interleavers. Its decoder exchanges soft information between the component decoders by using iterative method. The simulation to a simple example proves that compared with TTCM based on time invariant trellises with the same complexity, the proposed scheme has a better performance. A heuristic explanation based on maximizing effective free Euclidean distance is presented. We extend this time-varying scheme into higher bandwidth efficiency cases and compare it with traditional TTCM. Our scheme has similar performance and lower complexity.Thirdly, we discuss turbo equalization-the application of turbo principle on equalization. Because inter-symbol interference (ISI) in communication is similar to convolutional encoding, the concatenation of a convolutional code, an interleaver and ISI channel is similar to a serial concatenated convolutional code. So at the receiver we can equalize and decode jointly using iterative method as decoding a serial concatenated convolutional code. That is turbo equalization. We review the equivalent discrete time models of ISI channel and three kinds of traditional equalizer: Viterbi equalizer, linear equalizer and decision feedback equalizer. In order to apply these equalizers to turbo equalization, we present theirmodified versions using a priori information: MAP equalizer, CMMSE equalizer and DFE using a priori. We propose a modified CMMSE equalizer-RMMSE that uses an augmented real matrix representation for quadrature modulation systems. I...