Dynamic Sensor Sampling for IoT Edge Data Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Raw IoT data streams are often sent to the cloud in their entirety, which is impractical due to bandwidth limitations and the need for real-time processing, making it costly and inefficient, and requires intelligence at the edge for data management.
Innovation Solution
Implementing a system that dynamically adjusts the sampling interval of IoT sensor data streams based on events and business rules, allowing for real-time processing and sending only higher-level events to the cloud, using an IoT gateway with a complex event processing engine and rules management module to control data fidelity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If all raw IoT sensor data is sent to the cloud for processing, then data completeness is improved, but bandwidth usage and transmission costs increase significantly
Solution Approach 1:
The patent extracts and processes data locally at the edge device using machine learning models, sending only processed results or selectively filtered data to the cloud. This extraction of processing functionality from the cloud to the edge reduces the volume of data transmitted while maintaining data completeness for critical decisions.
Solution Approach 2:
The patent segments data processing into multiple levels: initial filtering at the sensor level, intermediate processing at the edge device using ML models, and final analysis at the cloud level. This segmentation allows each level to handle appropriate data granularity, reducing overall bandwidth usage while preserving essential information.
2Measurement precision
If all raw IoT sensor data is sent to the cloud for processing, then data accuracy is improved, but processing time and latency increase due to roundtrip communication
Solution Approach 1:
The patent performs preliminary data processing, filtering, and initial analysis at the edge device before cloud transmission. ML models pre-process sensor data locally to identify and transmit only significant events or anomalies, reducing cloud processing time while maintaining accuracy for critical decisions.
Solution Approach 2:
The edge device acts as an intermediary between sensors and the cloud, performing local ML-based processing to filter and prioritize data before cloud transmission. This intermediary processing reduces latency by handling time-sensitive operations locally while maintaining data accuracy through selective cloud validation.
3Quantity of substance
If all raw IoT sensor data is stored in cloud storage, then data retention is improved, but storage costs and bandwidth capacity requirements increase
Solution Approach 1:
The patent extracts valuable insights and processed information at the edge level, storing only essential processed data and results in the cloud rather than raw sensor streams. This extraction of processing functionality reduces cloud storage requirements while maintaining data retention for critical information.
Solution Approach 2:
The patent implements dynamic data retention strategies where the cloud stores processed results and aggregated data long-term, while edge devices maintain recent raw data buffers for immediate re-processing if needed. This dynamic allocation optimizes storage costs while ensuring data availability.
4Loss of energy
If edge devices perform intelligent processing of sensor data, then bandwidth usage is reduced, but device complexity increases
Solution Approach 1:
The patent implements self-service ML models at the edge that automatically adapt to sensor data patterns and processing requirements without extensive manual configuration. The models self-tune parameters and automatically filter data based on learned patterns, reducing operational complexity while maintaining intelligent processing capabilities.
Data Source
AI summary
A method includes receiving a sensor data stream from a sensor, determining that the sensor data stream is associated with a sensor fidelity rule specifying a triggering event, a fidelity modification, and a modification period, sampling the sensor data stream according to a first sampling interval, responsive to occurrence of a triggering event, changing the first sampling interval to a second sampling interval based on the fidelity modification, wherein the second sampling interval is different from the first sampling interval, determining whether the modification period has elapsed and whether the triggering event is active, and sampling the sensor data stream according to the second sampling interval when it is determined that the modification period has not elapsed and the triggering event is active.