Cascaded Video Analytics Edge Cloud Resource Allocation
Solution Overview
Problem
The increasing deployment of IoT cameras generates vast amounts of data that overburden network bandwidth and cloud processing capabilities, leading to inefficiencies and high computational costs in video analytics, with existing cloud-based systems often experiencing network unavailability and resource underutilization.
Innovation Solution
A dynamic video analytics system that allocates processing resources between edge devices and cloud devices based on available resources and threshold confidence values, using a cascading model to perform video analytics locally and shift processing to the cloud only when necessary, thereby reducing computational resource utilization and improving accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If all video content is processed at the cloud, then video analytics capability is provided, but computational cost and network cost increase significantly
Solution Approach 1:
The system segments video processing tasks between edge devices and cloud devices. Edge devices perform initial processing of video content locally, while only selected content requiring advanced analytics is transmitted to the cloud. This segmentation reduces network bandwidth consumption and cloud computational costs while maintaining video analytics capability.
Solution Approach 2:
Edge devices serve as intermediaries between cameras and cloud services. They receive video content from cameras, perform local processing and filtering, and selectively forward only necessary content to the cloud. This intermediary role reduces the volume of data transmitted over the network and the computational burden on cloud resources.
2Reliability
If cloud services are used for video analytics, then processing capability is provided, but network unavailability severely hinders system usefulness
Solution Approach 1:
The system extracts video processing capability from the cloud and places it at the edge devices. Edge devices can independently process video content locally without requiring cloud connectivity. This extraction enables the system to maintain video analytics functionality even when network connectivity to the cloud is unavailable.
Solution Approach 2:
The system dynamically adjusts its operation mode based on cloud availability. When the cloud is accessible, edge devices can offload complex analytics tasks to the cloud. When the cloud becomes unavailable, the system automatically transitions to local-only processing at edge devices, ensuring continuous operation without network dependency.
3Productivity
If processing is allocated between edge and cloud devices, then resource efficiency is improved, but system complexity increases
Solution Approach 1:
The system implements feedback mechanisms where edge devices monitor cloud availability, resource utilization, and video content characteristics. Based on this feedback, the system automatically adjusts the allocation of processing tasks between edge and cloud devices, optimizing resource efficiency without requiring complex manual configuration or centralized control.
Data Source
AI summary
This document relates to performing live video stream analytics on edge devices. One example determines resources available to the system, and a video analytics configuration is selected that distributes work between edge devices and cloud devices in a cascading manner, where edge device processing is prioritized over cloud processing in order to conserve resources. This example can dynamically modify the allocation of processing depending on changing conditions, such as network availability.