Feature Encoder for M2M Communications Reducing Data Redundancy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current M2M communication systems face inefficiencies due to redundancies in data transmission, particularly in massive machine-type communications, where multiple sensors observe the same subject, leading to inter-sensor, time-related, and machine-perception redundancies, and challenges in adapting to changing channel conditions and application requirements.
Innovation Solution
The use of trained encoder and decoder deep neural networks (DNNs) for feature-based compression, which extracts and transmits probabilistic features representing probability distributions, allowing for optimal compression ratios and sub-channel assignment based on feature importance, and cooperative management among sensors to reduce redundancies and optimize wireless transmission.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If multiple sensors transmit raw information about a common subject, then complete information is captured, but transmission redundancy increases and channel resources are wasted
Solution Approach 1:
The patent extracts only the essential features from raw sensor information rather than transmitting complete raw data. The encoder DNN identifies and transmits only the most important features that capture the essential information about the subject, eliminating redundant data transmission while preserving information completeness.
Solution Approach 2:
The system changes the parameter of information representation from raw data to compressed features with associated probability distributions. By transforming the information format and using variable-length coding based on feature importance, the system reduces transmission redundancy while maintaining information fidelity.
2Loss of energy
If compression is applied to reduce data transmission, then channel resource usage decreases, but information loss may occur
Solution Approach 1:
The system uses feedback loops where the encoder and decoder DNNs are trained together with the loss function incorporating both compression efficiency and information fidelity requirements. The training process continuously adjusts the encoding strategy to maintain acceptable information quality while maximizing compression ratios.
Solution Approach 2:
The compression ratio and feature selection are dynamically adjusted based on channel conditions and application requirements. The system can adaptively change which features to transmit and at what precision level, optimizing the balance between compression and fidelity for different operational scenarios.
3Loss of information
If all features are transmitted with equal importance, then information completeness is maintained, but transmission efficiency decreases
Solution Approach 1:
The patent applies different transmission quality levels to different features based on their importance. Critical features are transmitted with high precision and reliability, while less important features use lower precision or can be omitted entirely. This localized quality adjustment optimizes transmission efficiency while maintaining information completeness for essential attributes.
4Device complexity
If sensors operate independently without coordination, then system complexity is reduced, but redundancy among sensors increases
Solution Approach 1:
The patent merges the encoding operations of multiple sensors by training a single encoder DNN that processes information from all sensors. This unified encoder identifies redundant information across sensors and ensures that only non-redundant features are transmitted, reducing overall system redundancy while maintaining manageable complexity through centralized training.
Data Source
AI summary
Methods and apparatuses for feature-driven machine-to-machine communications are described. At a feature encoder, features are extracted from sensed raw information, to generate features that compress the raw information by a compression ratio. The feature encoder implements a probabilistic encoder to generate the features, each feature providing information about a respective probability distribution that each represents one or more aspects of the subject. The probabilistic encoder is designed to provide a compression ratio that satisfies a predetermined physical channel capacity limit for a transmission channel. The features are transmitted over the transmission channel.