Punctured vs Rateless Codes for Hybrid ARQ

Abstract

Two incremental redundancy hybrid ARQ (IR-code. Initially, only the systematic part of the codeword and HARQ) schemes are compared: one is based on LDPC code a selected number of parity bits are transmitted. The selected ensembles with random transmission assignments, the other parity bits together with the systematic bits form a codeword of is based on recently introduced Raptor codes. A number of a bits totherwtte sematic bis codeword important issues, such as rate and power control, and error rate a puctredemotherd. Deodngfissodis perfomed, performance after each transmission on time varying binary-at the receiving end. If a retransmission is requested, the input, symmetric-output channels are addressed by analyzing transmitter sends additional parity bits possibly under different performance of LDPC and Raptor codes on parallel channels. channel conditions or at different power. Decoding is again The theoretical results obtained for random code ensembles are attempted at the receiving end, where the new parity bits are tested on several practical code examples by simulation. Both theoretical and simulation results show that both LDPC and combined with those previously received. The procedure is Raptor codes are suitable for HARQ schemes. Which codes would repeated after each subsequent retransmission request until all make a better choice depends mainly on the width of the signal-the parity bits of the mother code are transmitted. to-noise operating range of the HARQ scheme, prior knowledge A historic overview of hybrid ARQ schemes up to 1998, can of that range, and other design parameters and constraints be found in [1]. Recent interest in the schemes comes from dictated by standards. the quest for reliable and efficient transmission under fluctuat-

Myths and realities of rateless coding

“…The discrete-input continuousoutput memoryless channel's (DCMC) capacity curve is also shown. The Raptor code and the punctured LDPC benchmarker codes follow the same designs presented in [14]. The decoder employs the well-known sum-product algorithm, limited to a maximum of 200 iterations.…”

Low-Density Parity-Check Codes and Their Rateless Relatives

“…Both methods have their own limitations and typically a combination of both techniques is generally preferred [233], [291]. The striking similarities of rateless coding with HARQ were first exploited by Soljanin et al in [292], [293], who compared the performance of Raptor codes as well as punctured LDPC codes for transmission over the BIAWGN channel. Their results demonstrated that the family of Raptor codes represents a more suitable alternative than punctured LDPCs for covering an extensive range of channel SNRs (and thus rates).…”

“…Average throughput (bits/channel use) performance for transmission over the BIAWGN channel versus SNR (dB) using the proposed reconfigurable rateless codes as well as for the Raptor code [254] and the incremental-redundancy-based HARQ schemes employing punctured regular LDPC codes having R = 0.8 and 0.9 and an optimized punctured half-rate irregular LDPC code. The Raptor code and the punctured LDPC benchmarker codes followed the design presented in [292], [293]. The decoder employed the SPA and was limited to a maximum of 200 iterations.…”

Amplify-forward and decode-forward cooperation relying on systematic Luby transform coded Hybrid Automatic-Repeat-reQuest

“…The performance of rateless codes, such as LT and Raptor codes, were compared with punctured codes in [17] and were further investigated in a wireless relay-aided network in [18]. Another coding scheme designed for cooperative communication is constituted by the HARQ scheme of [19,20].…”