Energy optimization of quantized min-sum decoders for protograph-based LDPC codes
- PDF / 464,896 Bytes
- 7 Pages / 595.224 x 790.955 pts Page_size
- 53 Downloads / 209 Views
Energy optimization of quantized min-sum decoders for protograph-based LDPC codes Mohamed Yaoumi1 · Elsa Dupraz1
· Franc¸ois Leduc-Primeau2 · Frederic Guilloud1
Received: 18 April 2020 / Accepted: 20 August 2020 © Institut Mines-T´el´ecom and Springer Nature Switzerland AG 2020
Abstract This paper considers protograph-based LDPC codes, and proposes an optimization method to select protographs that minimize the energy consumption of quantized min-sum decoders. This method first estimates the average number of iterations required by the decoder, and includes this estimate into two high-level models that evaluate the decoder energy consumption. The optimization problem is then formulated as minimizing the energy consumption of the decoder while satisfying a performance criterion on the frame error rate. Finally, an optimization algorithm based on differential evolution is introduced. Protograph optimized for energy consumption shows a gain in energy of approximately 15% compared with a baseline protograph optimized for performance only. Keywords LDPC codes · Protographs · Min-sum decoder · Energy
1 Introduction Reducing the energy consumption of telecommunication systems may allow improving the communication capabilities of systems with limited resources. This energy consumption comes for a large part from the transmission power of the emitter, but the processing power at the receiver is also non-negligible for short-length communications [1]. In addition, error-correction decoders are known to use a large part of this processing power. Therefore, the objective of this paper is to reduce the energy consumption of the error-correction part. It considers lowdensity parity check (LDPC) codes as a particular family Elsa Dupraz
[email protected] Mohamed Yaoumi [email protected] Franc¸ois Leduc-Primeau [email protected] Frederic Guilloud [email protected] 1
IMT Atlantique, Lab-STICC, UBL, Brest, France
2
Department of Electrical Engineering, Polytechnique Montreal, Montreal, QC, Canada
of error-correction codes which were retained in the 5G standardization process. The problem of reducing the energy consumption of LDPC decoders has received increased attention recently. In [2], the energy consumption of hard-decision LDPC decoders is estimated from the number of computation operations realized in the decoder, and from the average length of wires in the decoding circuit. Then, [3] considers more powerful LDPC decoders working on discrete message alphabets, and proposes a method to reduce the message alphabet size. This allows to lower both memory and wire energy consumption in the decoder. Alternatively, [4–6] propose to optimize the code parameters in order to reduce the decoder energy consumption. For instance, [4, 6] optimize the code degree distribution in order to minimize the decoder complexity. However, [6] considers hard-decision Gallager B decoders with poor performance, and [4] considers infinite precision sum-product decoding algorithms which
Data Loading...
