Time-Series Data Bucketing for Low-Latency Sensor Queries
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current database management systems face high latency and increased storage costs when performing temporal analytics on real-time sensor data, particularly due to the inefficiencies in processing and storage of time-sensitive data, which is exacerbated by the lack of compatibility with temporal normalization in parallel time series databases.
Innovation Solution
A computing system with a database server that receives data indexed by sensor identifiers and timestamps, creates time series data into buckets spaced at even intervals, and processes queries efficiently by locating target data within these buckets, thereby reducing latency and storage requirements through temporal normalization and massively parallel processing architecture.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If temporal normalization is performed using traditional SQL queries with OVERLAPS predicate, then data alignment and redundancy reduction are achieved, but processing latency increases substantially and storage space requirements increase
Solution Approach 1:
The patent segments the continuous time series data into discrete time buckets at predetermined intervals. This segmentation allows the database to store and query data in manageable chunks rather than processing entire continuous streams, dramatically reducing query latency while maintaining temporal normalization capabilities through bucket-based organization
Solution Approach 2:
The patent introduces a time bucket dimension to the data organization structure. By organizing data not only by sensor identifier but also by time bucket intervals, the system enables efficient temporal queries without requiring processor-intensive overlap predictions across entire data streams, thus reducing latency while maintaining alignment precision
2Productivity
If traditional database systems process real-time sensor data with temporal analytics, then comprehensive data analysis is achieved, but processing resources increase and latency becomes excessive
Solution Approach 1:
The system segments sensor data by time buckets and sensor identifiers, allowing parallel processing of discrete data segments. This segmentation enables the database to process only relevant time intervals and sensors rather than analyzing entire data streams, reducing processing resource consumption while maintaining comprehensive temporal analytics capability
Solution Approach 2:
The patent implements partial action by allowing queries to retrieve data for specific time buckets and sensors rather than requiring complete data analysis. This selective querying approach reduces processing resources by focusing only on the necessary subset of data while still achieving the required analytical insights
3Loss of energy
If parallel time series databases are used for time-stamped data, then data retrieval cost is reduced, but compatibility with temporal normalization is lost due to multiplying storage space and retrieval cost
Solution Approach 1:
The patent creates a universal data organization structure that combines the benefits of parallel time series databases with temporal normalization capabilities. By organizing data into time buckets that can be independently queried and processed, the system achieves both low retrieval costs through parallel processing and temporal normalization compatibility through structured time-based organization
Solution Approach 2:
The introduction of time bucket intervals as an additional organizational dimension enables the database to simultaneously support efficient parallel retrieval operations and temporal normalization queries. This dimensional organization allows the system to maintain low retrieval costs while gaining adaptability for temporal analytics
Data Source
AI summary
A computing system includes a database server configured to receive data including a plurality of elements each including a sensor identifier and time element, each time element including a timestamp or time period, and create a time series of the data indexed by the sensor identifiers into buckets spaced at even time intervals. The database server is configured to receive a query including at least one of the sensor identifiers and at least one query time element. The database server is configured to locate target data in the buckets by the at least one of the sensor identifiers and the at least one query time element and output the target data.