Sensor Processing System Using Domain Transform for Substream Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional data storage and computation systems face inefficiencies in processing large amounts of sensor data, particularly in terms of cost and power consumption when performing computations on stored data or analyzing real-time data streams, and there is a need for a more efficient way to handle bandwidth-heavy computations and data storage in distributed systems.
Innovation Solution
A sensor processing system that utilizes domain transforms to split digital sensor streams into reduced-size substreams, where the larger substream is stored in cold storage and the smaller substream is processed using deep machine learning on storage compute devices with faster access times, allowing for reduced computational and storage requirements and enabling efficient edge-based processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If deep machine learning is performed on the entire digital sensor stream, then analysis accuracy is improved, but computational resources and power consumption increase significantly
Solution Approach 1:
The patent applies segmentation by dividing the digital sensor stream into two substreams using domain transform: a first substream containing most of the data (stored in cold storage) and a second substream containing a smaller portion (processed with deep machine learning). This segmentation allows the system to perform computationally intensive deep machine learning only on the smaller second substream while maintaining analysis accuracy, thereby significantly reducing power consumption compared to processing the entire sensor stream.
2Measurement precision
If deep machine learning is performed on the entire digital sensor stream, then analysis accuracy is improved, but computational resources increase significantly
Solution Approach 1:
The patent segments the digital sensor stream into two substreams through domain transform, where only the smaller second substream undergoes deep machine learning processing. This segmentation reduces the computational burden and resource requirements while maintaining analysis accuracy by preserving critical information in the reduced substream.
Solution Approach 2:
The patent extracts the essential information needed for accurate analysis into a smaller second substream through domain transform. By taking out only the critical portion of the data that contains the necessary patterns for machine learning analysis, the system reduces computational resource requirements while maintaining analysis accuracy.
3Speed
If the entire digital sensor stream is stored in fast storage, then data access speed is improved, but storage costs and power consumption increase
Solution Approach 1:
The patent segments the digital sensor stream into two substreams with different storage requirements. The first substream, containing the majority of the data, is stored in cold storage (slower, lower power). The second substream, containing critical information for analysis, is stored in fast storage for quick access during machine learning processing. This segmentation enables the system to maintain fast access speed for essential data while reducing overall power consumption by storing less critical data in lower-power storage.
4Speed
If the entire digital sensor stream is stored in fast storage, then data access speed is improved, but storage costs increase
Solution Approach 1:
The patent segments the digital sensor stream into two substreams stored in different storage tiers. The first substream is stored in cost-effective cold storage, while the smaller second substream requiring frequent access for analysis is stored in expensive fast storage. This segmentation allows the system to maintain fast access speed for critical data while significantly reducing storage costs by storing the majority of data in cheaper storage media.
Data Source
AI summary
A digital sensor stream is received from a sensor. A domain transform is performed on the digital sensor stream to produce first and second substreams. The first substream is larger than the second substream. The first substream is stored in cold storage and the second substream is stored in a second storage that has faster access times than the cold storage. Deep machine learning is performed on the second substream, the results of which may be stored in the second storage.