Edge Sensor Data Processing Accuracy via Cross-Device Training
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current smart-city systems using Cloud computing architectures face limitations in data flow and accuracy due to their centralized nature, which hampers the effectiveness of edge computing in processing environmental data from sensors.
Innovation Solution
A system comprising edge devices with multiple sensors and processing means that share knowledge and rules to improve data processing accuracy, where one sensor can teach another how to enhance its data processing by using processed data from another sensor, enabling more accurate and comprehensive data analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If Cloud computing architecture is used for smart-city systems, then centralized data processing is achieved, but data flow freedom and processing accuracy are limited
Solution Approach 1:
The patent segments the centralized cloud computing architecture into distributed edge computing nodes. Each edge device independently processes sensor data using local processing means, eliminating the single-point bottleneck of centralized cloud architecture. This segmentation enables parallel processing across multiple devices, improving both data flow freedom and processing accuracy while reducing architectural complexity.
Solution Approach 2:
The patent introduces a new dimension of processing by implementing cross-device knowledge transfer. Instead of single-device independent processing, the system enables processing means of one device to be trained using processed data from sensors of other devices. This dimensional expansion from isolated to collaborative processing enhances accuracy without increasing centralized complexity.
2Productivity
If edge computing is implemented to process data at sensor level, then data flow freedom improves, but sensing quality becomes a limiting factor
Solution Approach 1:
The patent merges the capabilities of multiple sensors across different edge devices by enabling knowledge sharing. The processing means of one device learns from the processed data of another device's sensors, effectively combining their sensing capabilities. This merging allows the system to overcome the limitations of individual sensor quality while maintaining distributed edge computing efficiency.
Solution Approach 2:
The patent implements a copying mechanism where processed data from one device serves as training data for another device's processing means. The knowledge gained from one sensor's processed data is copied and applied to improve another sensor's processing accuracy, enabling quality enhancement without requiring additional physical sensors.
3Loss of information
If multiple sensors are used to observe events, then data comprehensiveness improves, but system complexity increases
Solution Approach 1:
The patent makes the processing means universal by enabling them to process data from multiple different sensor types. Instead of requiring specialized processing logic for each sensor, the processing means can be trained on processed data from various sensors and apply this knowledge universally. This multi-functionality reduces complexity while maintaining comprehensive information processing across multiple sensors.
Data Source
AI summary
A system comprising: at least a first sensor (13a) and a second sensor (13b) arranged at one or more edge devices (10), the first sensor (13a), respectively the second sensor (13b), being configured for obtaining first, respectively second, environmental data (Eda, Edb) related to an 5 event in the vicinity of the one or more edge devices; a first processing means (12a, 22a) configured to process said first environmental data (EDa) in accordance with a first set of rules to generate first processed data (PDa) and a second processing means (12b, 22b) configured to process said second environmental data (EDb) in accordance with a second set of rules to generate second processed data (PDb); a control means (17, 27) configured to control the first and second 10 processing means such that the second processed data (PD2) is used to train the first processing means.