Mesh Network Packet Decoding Using Grouped Likelihood Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In meshed communications networks, the decoding of LDPC-encoded data packets is impaired by cycles in the parity check matrix, leading to suboptimal performance due to correlations between rows and columns, especially when packets result from the same linear combination, causing increased bit error rates.
Innovation Solution
A method involving two consecutive decoding operations at the destination node: the first decoding is performed by groups of packets using a belief propagation algorithm to exploit packet redundancies, and the second decoding combines likelihood information from these groups, utilizing smaller-sized decoding matrices and averaging trust information to reduce bit error rates and optimize bandwidth.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If LDPC decoding is performed using all received packets including combined packets from the same linear combination, then more packets are utilized for decoding, but cycles appear in the parity check matrix causing correlation between rows and columns which increases bit error rates
Solution Approach 1:
The patent segments the received packets into distinct groups based on their linear combination origins. Each group contains packets that share the same combination coefficients, allowing the decoder to process them separately. This segmentation prevents the formation of cycles in the parity check matrix by ensuring that each packet group contributes independently to the decoding process, thereby maintaining row and column independence while still utilizing all available packets for decoding.
2Loss of energy
If combined packets from the same linear combination are used in decoding, then bandwidth utilization improves, but the correlation between packet rows in the parity matrix degrades decoding performance
Solution Approach 1:
The patent applies preliminary action by pre-processing the received packets to identify and categorize them into groups based on their linear combination characteristics before the actual decoding process. This preliminary classification allows the decoder to construct parity check matrices that avoid correlations, ensuring optimal decoding accuracy while maintaining efficient bandwidth utilization from the combined packets.
3Ease of operation
If standard LDPC decoding is applied to packets from meshed networks with relay combinations, then implementation remains simple, but performance is suboptimal due to matrix cycles
Solution Approach 1:
The patent introduces dynamics by making the decoding process adaptive to the specific structure of received packets from meshed networks. The decoder dynamically adjusts its operation by detecting packet groups with identical combination coefficients and processing them through specialized sub-routines that prevent cycle formation. This dynamic approach maintains implementation simplicity while significantly improving decoding performance in meshed network environments.
Data Source
AI summary
A method is proposed for decoding a plurality of data packets received through a meshed communications network for the retrieval of source packets transmitted by one or more source nodes, the meshed communications network comprising relay nodes generating combined packets, each combined packet consisting of a linear combination of source packets. During the decoding by a destination node, this method consists in performing two decoding operations of which the first is a decoding by group of packets and the second is a decoding that takes account of pieces of likelihood information resulting from the first decoding. The fact of first of all carrying out a decoding with groups of packets makes it possible to exploit the repetitions (or redundancies) of packets in a meshed network to optimize the bit error rate during the decoding of source data transmitted on the communications network.