Wang, Guan (2010) Hardware implementation of a pipelined turbo decoder. Masters thesis, Memorial University of Newfoundland.
- Accepted Version
Available under License - The author retains copyright ownership and moral rights in this thesis. Neither the thesis nor substantial extracts from it may be printed or otherwise reproduced without the author's permission.
Turbo codes have been widely studied since they were first proposed in 1993 by Berrou, Glavieux, and Thitimajshima in "Near Shannon Limit error-correcting coding and decoding: Turbo-codes" . They have the advantage of providing a low bit error rate (BER) in decoding, and outperform linear block and convolutional codes in low signal-to-noise-ratio (SNR) environments. The decoding performance of turbo codes can be very close to the Shannon Limit, about 0.7decibel (dB). It is determined by the architectures of the constituent encoders and interleaver, but is bounded in high SNRs by an error floor. Turbo codes are widely used in communications. We explore the codeword weight spectrum properties that contribute to their excellent performance. Furthermore, the decoding performance is analyzed and compared with the free distance asymptotic performance. A 16-state turbo decoder is implemented using VHSIC Hardware Description Language (VHDL) and then mapped onto a field-programmable gate array (FPGA) board. The hardware implementations are compared with the software simulations to verify the decoding correctness. A pipelined architecture is then implemented which significantly reduces the decoding latency. -- Keywords: turbo codes; decoding performance; Monte Carlo simulations; FPGA implementation
|Item Type:||Thesis (Masters)|
|Additional Information:||Includes bibliographical references (leaves 91-94).|
|Department(s):||Engineering and Applied Science, Faculty of|
|Library of Congress Subject Heading:||Decoders (Electronics)--Design and construction; Error-correcting codes (Information theory); Monte Carlo method; VHDL (Computer hardware description language)|
Actions (login required)