Deep-Learning-Based Adaptive Error-Correction Decoding for Spin-Torque Transfer Magnetic Random Access Memory (STT-MRAM)

TL;DR

The paper tackles reliable data recovery in STT-MRAM under die-to-die BER variation and unknown temperature-induced resistance offsets by introducing a unified neural decoder framework. It develops Neural BF (NBF), which shares a trellis with neural belief propagation (NBP) and neural offset min-sum (NOMS), enabling a single architecture to realize multiple decoding strategies. A DL-based adaptive decoding algorithm selects decoding level based on channel BER without channel knowledge, reducing latency and energy. Experimental results show neural decoders improve BER over standard decoders with similar latency, and the adaptive scheme achieves about a 50% reduction in latency and energy compared to a fixed high-complexity decoder, while remaining robust to unknown offsets.

Abstract

Spin-torque transfer magnetic random access memory (STT-MRAM) is a promising emerging non-volatile memory (NVM) technology with wide applications. However, the data recovery of STT-MRAM is affected by the diversity of channel raw bit error rate (BER) across different dies caused by process variations, as well as the unknown resistance offset due to temperature change. Therefore, it is critical to develop effective decoding algorithms of error correction codes (ECCs) for STT-MRAM. In this article, we first propose a neural bit-flipping (BF) decoding algorithm, which can share the same trellis representation as the state-of-the-art neural decoding algorithms, such as the neural belief propagation (NBP) and neural offset min-sum (NOMS) algorithm. Hence, a neural network (NN) decoder with a uniform architecture but different NN parameters can realize all these neural decoding algorithms. Based on such a unified NN decoder architecture, we further propose a novel deep-learning (DL)-based adaptive decoding algorithm whose decoding complexity can be adjusted according to the change of the channel conditions of STT-MRAM. Extensive experimental evaluation results demonstrate that the proposed neural decoders can greatly improve the performance over the standard decoders, with similar decoding latency and energy consumption. Moreover, the DL-based adaptive decoder can work well over different channel conditions of STT-MRAM irrespective of the unknown resistance offset, with a 50% reduction of the decoding latency and energy consumption compared to the fixed decoder.

Figures (6)

• Figure 1: The probability density functions (PDFs) of resistances for STT-MRAM: $R_{1}$ represents the initial high resistance, and $R_{1}^{'}$ is the shifted high resistance caused by the drop of temperature, and $R_{0}$ represents the fixed low resistance.
• Figure 2: DNN architecture for the (7,4) Hamming code with 4 hidden layers corresponding to 2 full decoding iterations.
• Figure 3: Block diagram of the proposed DL-based adaptive decoding algorithm.
• Figure 4: Total cycles of different decoders with $I=5$.
• Figure 5: Energy consumption of different decoders with $I=5$.