Neural Network Interleaver for Variable-Entropy Wireless Decoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional interleavers in wireless communication systems assume constant source entropy for control bits, failing to account for varying entropies due to UE patterns, leading to suboptimal decoding performance and increased block error rates.
Innovation Solution
Employ a neural network (NN) to generate an adaptive interleaver based on channel and UE connection status, using a shared NN to interleave and deinterleave bits, optimizing the interleaver for diverse entropies and improving reliability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a conventional interleaver is used with constant source entropy assumption, then the device complexity is low, but the reliability deteriorates due to suboptimal decoding performance
Solution Approach 1:
The patent applies dynamics by transitioning from a static interleaver design with fixed permutation patterns to a dynamic interleaver generation process using neural networks. The NN adapts the interleaver configuration based on varying channel conditions and traffic patterns, allowing the system to optimize decoding performance for different entropy scenarios while maintaining manageable complexity through learned patterns.
Solution Approach 2:
The patent changes the parameter of source entropy from a constant assumption to a variable parameter that the neural network can detect and adapt to. By using NN to estimate actual source entropy and generate corresponding interleaver patterns, the system achieves improved reliability without proportionally increasing complexity, as the NN learns efficient parameter mappings during training.
2Reliability
If an adaptive interleaver is designed to account for varying source entropy, then the reliability improves, but the device complexity increases
Solution Approach 1:
The patent uses copying by training the neural network on extensive simulation data and channel models beforehand. The NN learns to replicate the behavior of complex optimal interleavers for various channel conditions, allowing the actual system to use the trained NN (a simplified copy) rather than implementing the full complex optimization algorithms in real-time, thus improving reliability while controlling complexity.
Solution Approach 2:
The patent applies preliminary action by pre-training the neural network with extensive channel models and traffic patterns before deployment. This offline training phase performs the complex adaptive work in advance, allowing the deployed system to achieve high reliability through the pre-trained NN without bearing the full computational burden during actual communication operations.
3Reliability
If the interleaver is optimized for specific channel conditions, then the reliability improves, but the adaptability deteriorates
Solution Approach 1:
The patent achieves universality by designing a single neural network that can handle multiple channel conditions and traffic patterns. The NN is trained on diverse channel models and entropy scenarios, enabling it to universally adapt to various conditions while maintaining optimized decoding reliability for each specific scenario, eliminating the need for separate interleavers for different conditions.
Solution Approach 2:
The patent implements feedback by using the neural network to continuously adapt the interleaver based on detected channel conditions and traffic patterns. The system monitors actual performance and channel state, feeds this information to the NN, which then adjusts the interleaver configuration in real-time, maintaining both reliability and adaptability through closed-loop control.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present disclosure relates to a technical solution that improves the reliability of communications over a wireless communication channel by replacing a conventional interleaver (e.g., random interleaver) with a Neural Network (NN)-generated interleaver. For this purpose, a well-trained NN is used, which is configured to receive a UE connection status and a channel status as input data and outputs the interleaver in the form of an orthogonal binary matrix. The NN is shared by a UE and a network node. The network node may use the interleaver to interleave a set of bits in a downlink control message before the downlink control message is encoded, e.g., with an error correcting code, such as a polar code. The UE may generate and transpose the interleaver to obtain a deinterleaver to be applied to the downlink control message after its decoding (e.g., polar decoding).