Recurrent Neural Network Decoding for Low-Latency ECC
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing error correction coding techniques in memory devices and wireless baseband circuitry increase processing complexity and resource usage, leading to inefficiencies in frequency, channel, and storage resource usage, particularly in applications requiring ultra-low power consumption and latency like IoT and tactile internet.
Innovation Solution
The use of multi-layer neural networks and recurrent neural networks for decoding encoded data, which leverage nonlinear mapping and distributed processing capabilities to reduce errors introduced by noise, such as bit flips in non-volatile memory devices, by transforming noisy encoded input data into decoded data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If complex error correction coding techniques are used to reduce errors in noisy environments, then reliability is improved, but device complexity and processing resources increase
Solution Approach 1:
The patent applies preliminary action by performing offline training of neural network decoders before actual decoding operations. During offline training, the neural network learns optimal decoding strategies for various error correction codes, storing pre-computed weight parameters. This pre-processing transfers complexity from real-time operation to offline preparation, reducing runtime processing requirements while maintaining high reliability in error correction.
2Reliability
If complex error correction coding techniques are used to reduce errors in noisy environments, then reliability is improved, but processing speed decreases
Solution Approach 1:
The patent applies preliminary action by performing offline training of neural network decoders before actual decoding operations. During offline training, the neural network learns optimal decoding strategies for various error correction codes, storing pre-computed weight parameters. This pre-processing transfers complexity from real-time operation to offline preparation, reducing runtime processing requirements while maintaining high reliability in error correction.
3Device complexity
If traditional decoding methods are used to decode encoded data, then device complexity is reduced, but error reduction capability in noisy environments deteriorates
Solution Approach 1:
The patent applies mechanics substitution by replacing traditional mechanical/error-prone decoding algorithms with neural network-based decoding. The neural network decoder uses learned weight parameters and nonlinear transformations to achieve superior error correction performance compared to conventional algebraic decoding methods, while the offline training phase prepares the network to operate efficiently during runtime.
4Productivity
If offline training of neural networks is performed to transfer complexity, then processing speed during decoding is improved, but loss of time during training increases
Solution Approach 1:
The patent applies preliminary action by performing offline training of neural network decoders before actual decoding operations. During offline training, the neural network learns optimal decoding strategies for various error correction codes, storing pre-computed weight parameters. This pre-processing transfers complexity from real-time operation to offline preparation, reducing runtime processing requirements while maintaining high reliability in error correction.
Data Source
AI summary
Examples described herein utilize multi-layer neural networks, such as multi-layer recurrent neural networks to decode encoded data (e.g., data encoded using one or more encoding techniques). The neural networks and/or recurrent neural networks have nonlinear mapping and distributed processing capabilities which are advantageous in many systems employing the neural network decoders and/or recurrent neural networks. In this manner, neural networks or recurrent neural networks described herein are used to implement error correction coding (ECC) decoders.