Presentation | 2021-06-25 A Study on Deep Unfolding-Aided Quantized AMP for High-Dimensional Signal Detection Atsunori Shimamura, Takumi Takahashi, Shinsuke Ibi, Seiichi Sampei, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | We consider a look-up table (LUT)-based multi-user detection (MUD) via quantized approximate message passing (AMP) in large multi-user multi-input multi-output (MU-MIMO) systems. Although AMP is well known as low-complexity MUD, when internal operations are conducted with double-precision arithmetic, large memory occupancy and severe processing delay are inevitable in the actual large-scale implementation. To address this issue, the LUT-based MUD replaces all operations with a simple LUT search and integer arithmetic, where all messages exchanged between each iteration process are unsigned integers. This paper proposes a deep unfolding-aided approach to minimize the negative effects of quantization errors throughout the iterative process via data-driven tuning of quantization thresholds in the LUT, where the non-differentilable LUT operations are approximately replaced with soft threshold functions, enabling the tuning of the quantization thresholds via the stochastic gradient descent (SGD) and backpropagation. Numerical results demonstrate the efficacy of the proposed method in terms of bit error rate (BER) performance and memory usage. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | Large MIMO detection / quantization / look-up table / deep unfolding / belief propagation |
Paper # | RCS2021-68 |
Date of Issue | 2021-06-16 (RCS) |
Conference Information | |
Committee | RCS |
---|---|
Conference Date | 2021/6/23(3days) |
Place (in Japanese) | (See Japanese page) |
Place (in English) | Online |
Topics (in Japanese) | (See Japanese page) |
Topics (in English) | First Presentation in IEICE Technical Committee, Resource Control, Scheduling, Wireless Communications, etc. |
Chair | Eiji Okamoto(Nagoya Inst. of Tech.) |
Vice Chair | Fumihide Kojima(NICT) / Toshihiko Nishimura(Hokkaido Univ.) / Tomoya Tandai(Toshiba) |
Secretary | Fumihide Kojima(Panasonic) / Toshihiko Nishimura(NEC) / Tomoya Tandai |
Assistant | Koichi Adachi(Univ. of Electro-Comm.) / Osamu Nakamura(Sharp) / Manabu Sakai(Mitsubishi Electric) / Masashi Iwabuchi(NTT) / Tatsuki Okuyama(NTT DOCOMO) |
Paper Information | |
Registration To | Technical Committee on Radio Communication Systems |
---|---|
Language | JPN |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | A Study on Deep Unfolding-Aided Quantized AMP for High-Dimensional Signal Detection |
Sub Title (in English) | |
Keyword(1) | Large MIMO detection |
Keyword(2) | quantization |
Keyword(3) | look-up table |
Keyword(4) | deep unfolding |
Keyword(5) | belief propagation |
1st Author's Name | Atsunori Shimamura |
1st Author's Affiliation | Osaka University(Osaka Univ) |
2nd Author's Name | Takumi Takahashi |
2nd Author's Affiliation | Osaka University(Osaka Univ) |
3rd Author's Name | Shinsuke Ibi |
3rd Author's Affiliation | Doshisha University(Doshisha Univ) |
4th Author's Name | Seiichi Sampei |
4th Author's Affiliation | Osaka University(Osaka Univ) |
Date | 2021-06-25 |
Paper # | RCS2021-68 |
Volume (vol) | vol.121 |
Number (no) | RCS-72 |
Page | pp.pp.226-231(RCS), |
#Pages | 6 |
Date of Issue | 2021-06-16 (RCS) |