Design Of A Viterbi Decoder Compatible With DVB-T Standards

Error Correct Coding (ECC) and the Error-Control technology have played more and more important role in our life among which the convolutional code is adopted widely because of its excellent capability. The Viterbi algorithm is a maximum likelihood decoding procedure for the convolutional code. It has developed rapidly both in theory and in practice since the day it was presented. This paper does a research on the Viterbi decoder which is compatible with DVB-T standards. Viterbi decoder is the critical part of the inner-code decoding of the DVB-T receiver with important research value.In this thesis, on the basis of the decoding algorithms and their relative technologies developed in these fields, we realize our design in behavior level from the point of reducing the complexity and power dissipation independtly. The Viterbi decoder has been implemented in xc2v1000 device of Xilinx VirtexII series successfully.These works are presented in this thesis:1. In our case soft decision decoding is used so the input symbol is represented as 3-bit offset code. Several mostly used algorithms to compute the Euclidean distance are analyzed. The process is simplified in the theory so that it can be implemented in hardware more easily.2. The process of depuncturing is discussed in detail. The traditional architecture with FIFO is abounded to save the resource and reduce the complexity in hardware implementation.3. In order to reduce the power dissipation the modified T-algorithm is used to implement the add-compare-select units so that the path metric can be represented in 5 bits instead of 9 bits of the traditional method. The area-efficient architecture is used in the path metric management unit to further reduce the complexity compared with the full-parallel structure. The pre-compute method is used to get the new path metric value. In this way the power can be reduced by 30% compare to the traditional architecture.4. In order to save the memory source we use the in-place update method to manage the path metric. Modified Hybrid Trace Forward architecture with lower memory source is presented which has low decoder latency and power dissipation. According to the character of the (2, 1, 7) convolutional code we specify the length...