How to Boost the Throughput of HARQ With Off-the-Shelf Codes

Abstract

In this work, we propose a coding strategy designed to enhance the throughput of hybrid ARQ (HARQ) transmissions over i.i.d. block-fading channels with the channel state information (CSI) unknown at the transmitter. We use a joint packet coding where the same channel block is logically shared among many packets. To reduce the complexity, we use a two-layer coding where, first, packets are first coded by the binary compressing encoders, and the results are then passed to the conventional channel encoder. We show how to optimize the compression rates on the basis of the empirical error-rate curves.We also discuss how the parameters of the practical turbo-codes may be modified to take advantage of the proposed HARQ scheme. Finally, simple and pragmatic rate adaptation strategies are developed.In numerical examples, our scheme is compared to the conventional incremental redundancy HARQ (IR-HARQ), and it yields a notable gain of 1 − 2dB in the region of high throughput, where HARQ fails to provide any improvement.

AMC and HARQ: How to Increase the Throughput

“…2 Indeed, the literature already recognizes that, to improve the throughput, the coding across the HARQ rounds must be modified. The most relevant solutions may be classified as i) a multi-packet coding [2], [3], [20], [21], where many packets with variable contents are jointly encoded into fixed-length codewords which then use the fixed resources (channel blocks) or, as ii) a variable-length coding [22]- [27], where rather the codewords length varies throughout the HARQ rounds and the packet content is fixed. 3 We focus this work on the adaptive multi-packet coding whose advantage over the relatively well-studied variablelength coding will be discussed in Sec.…”

“…To make the multi-packet approach practical, we will use L-HARQ, studied in [2], [3] and implemented via a two-step (or, layered) approach, where the joint coding is implemented in two independent steps: the binary packet mixing is followed by the conventional channel coding. From the theoretical perspective, when compared to a more general joint coding/decoding scheme [21], [30], L-HARQ does not impose any throughput penalty [30,Th.…”

“…We formulate the optimization problem and propose a framework to solve it. Concluding on the excessive numerical complexity of the latter, we present heuristic rate-adaptation policies inspired by the optimal adaptation policy used in the AMC and by the results of [3]. The proposed solution leads to a local adaptation with easily adjustable parameters and does not require the knowledge of the entire model of the relationship between the different HARQ rounds, the AMC, and the fading model.…”

“…, K to enable the decoding of the packet m 1 , transmitted in the first round, we rather transmit (some of) the bits of the packet m 1 (we denote these redundant bits as m ′ 1 ), and we "fill" the remaining space with a fresh content m 2 . So the conventional channel coding is preceded by the mixing of the punctured information packets; this is the idea of layeredcoding, L-HARQ, proposed in [2] and modified in [3]. The decoding is also done in layers, where the mixed packets m ′ 1 and m 2 are decoded first and, next, the packet m 1 is recovered with the aid of m ′ 1 .…”

“…The decoding is also done in layers, where the mixed packets m ′ 1 and m 2 are decoded first and, next, the packet m 1 is recovered with the aid of m ′ 1 . The main advantages are that i) the channel coding is not affected by the packet mixing and can be implemented without joint-coding considerations, ii) the decoding is straightforward and can be done with commercially available decoders as demonstrated in [3], and finally iii) the decoding results are described by one-dimensional PER curves which depend only on one SNR; the dependence of the decoding results on other SNRs is implicit and defined via recursive (layerer) decoding.…”