Abstract
In smart city scenarios, the huge proliferation of monitoring cameras scattered in public spaces has posed many challenges to network and processing infrastructure. A few dozen cameras are enough to saturate the city’s backbone. In addition, most smart city applications require a real-time response from the system in charge of processing such large-scale video streams. Finding a missing person using facial recognition technology is one of these applications that require immediate action on the place where that person is. In this paper, we tackle these challenges presenting a distributed system for video analytics designed to leverage edge computing capabilities. Our approach encompasses architecture, methods, and algorithms for: (i) dividing the burdensome processing of large-scale video streams into various machine learning tasks; and (ii) deploying these tasks as a workflow of data processing in edge devices equipped with hardware accelerators for neural networks. We also propose the reuse of nodes running tasks shared by multiple applications, e.g., facial recognition, thus improving the system’s processing throughput. Simulations showed that, with our algorithm to distribute the workload, the time to process a workflow is about 33% faster than a naive approach.
Highlights
Since the advent of the Internet of Things (IoT) in the early 2000s [1], the Internet has become the central infrastructure for transporting all types of information, from simple text to complex multimedia streams, produced by both physical and virtual entities
We developed an operator that encapsulates the task program/source code and additional files, such as the deep learning (DL) model file used in the inference processes for Measurement Level Tasks (MLT)
We performed four simulations varying the number of feature level operator (FLO) nodes and the generation rate of images of interest yielded by measurement level operator (MLO) nodes when processing the video frames
Summary
Since the advent of the Internet of Things (IoT) in the early 2000s [1], the Internet has become the central infrastructure for transporting all types of information, from simple text to complex multimedia streams, produced by both physical and virtual entities. On the one hand, processing at the network edge reduces bandwidth consumption and latency, on the other hand, edge computing resources are limited and inferior compared to those found in cloud data centers In this context, the following issue arises: how to process large-scale video streams to yield real-time events of interest using edge computing while meeting high-throughput criteria. The following issue arises: how to process large-scale video streams to yield real-time events of interest using edge computing while meeting high-throughput criteria To address this issue, it is necessary to propose a new strategy to distribute and allocate resources in edge nodes in an efficient and fair way, while meeting the applications’ requirements.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
