References & Citations
Computer Science > Information Theory
Title: GRAND for Rayleigh Fading Channels
(Submitted on 29 Apr 2022 (v1), last revised 1 Dec 2022 (this version, v2))
Abstract: Guessing Random Additive Noise Decoding (GRAND) is a code-agnostic decoding technique for short-length and high-rate channel codes. GRAND tries to guess the channel noise by generating test error patterns (TEPs), and the sequence of the TEPs is the main difference between different GRAND variants. In this work, we extend the application of GRAND to multipath frequency non-selective Rayleigh fading communication channels, and we refer to this GRAND variant as Fading-GRAND. The proposed Fading-GRAND adapts its TEP generation to the fading conditions of the underlying communication channel, outperforming traditional channel code decoders in scenarios with $L$ spatial diversity branches as well as scenarios with no diversity. Numerical simulation results show that the Fading-GRAND outperforms the traditional Berlekamp-Massey (B-M) decoder for decoding BCH code $(127,106)$ and BCH code $(127,113)$ by $\mathbf{0.5\sim6.5}$ dB at a target FER of $10^{-7}$. Similarly, Fading-GRAND outperforms GRANDAB, the hard-input variation of GRAND, by $0.2\sim8$ dB at a target FER of $10^{-7}$ with CRC $(128,104)$ code and RLC $(128,104)$. Furthermore the average complexity of Fading-GRAND, at $\frac{E_b}{N_0}$ corresponding to target FER of $10^{-7}$, is $\frac{1}{2}\times\sim \frac{1}{46}\times$ the complexity of GRANDAB.
Submission history
From: Syed Mohsin Abbas Dr. [view email][v1] Fri, 29 Apr 2022 18:22:06 GMT (568kb,D)
[v2] Thu, 1 Dec 2022 03:50:37 GMT (846kb,D)
Link back to: arXiv, form interface, contact.