Study of Hybrid Automatic Request Schemes and Tree-Search Decoding of Linear Block Codes

博士 === 國立臺灣大學 === 電信工程學研究所 === 101 === In this dissertation, we study two topics on communications: hybrid automatic repeat request (ARQ) and tree-search decoding of linear block codes. For the first topic, we propose an hybrid ARQ (HARQ) scheme with subpacket transmission and subpacket scheduling...

Full description

Bibliographic Details
Main Authors: Tien-Yu Lin, 林典育
Other Authors: Mao-Chao Lin
Format: Others
Language:en_US
Published: 2013
Online Access:http://ndltd.ncl.edu.tw/handle/71231770591464490295
Description
Summary:博士 === 國立臺灣大學 === 電信工程學研究所 === 101 === In this dissertation, we study two topics on communications: hybrid automatic repeat request (ARQ) and tree-search decoding of linear block codes. For the first topic, we propose an hybrid ARQ (HARQ) scheme with subpacket transmission and subpacket scheduling for systems with constant packet lengths. In this scheme, each transmission packet comprises two subpackets of equal lengths. As compared to the conventional type II HARQ scheme, the proposed HARQ scheme can more effectively control the error-correcting capability and thus attain better throughput efficiency in the moderate to high signal-to-noise ratios (SNR) regime. Based on the proposed HARQ scheme, we then presents two modified versions of the proposed scheme. The first version with higher complexity, can provide further throughput improvement in low SNR. The second version, with the highest complexity, can obtain additional throughput than the first version in moderate to high SNR. For the second topic, we investigate a tree-search decoding algorithm, named A*, and aim to effectively reduce the decoding latency (complexity). We propose two complexity-reduction techniques for the A* algorithm. The first technique is that the searching is embedded with depth constraints in which the numbers of bit difference from the most reliable positions at different depths are limited. In the second technique, apart from the tree searching, the algorithm employs generation of candidate codewords based on processing the newly updated candidate codeword. For both proposed algorithms, the searching complexity can be significantly reduced at the cost of slight performance loss. In addition, the two proposed techniques can be effectively combined to obtain a more efficient modified A* algorithm. We also investigate the A* decoding for block coded schemes with interblock memory. With interblock memory, both the error performance and average decoding latency of the A* algorithm can be significantly improved in the high SNR regime. Finally, we discuss a HARQ scheme which employs packet division together with the A* decoding. This scheme is easy to be implemented and can perform as well throughput as the conventional type II HARQ scheme.