资源限制削弱了边缘物联网的投资回报率(ROI)

温馨提示:全文约2019字,阅读全文大约需要3分钟

当网络边缘的物联网继续取得进展时,资源限制给这些设备带来了巨大的挑战

.

物联网设备如今受益于边缘较低的网络延迟和设备上更高的数据智能。边缘处理绕过了集中式计算的时间延迟和数据安全挑战:数据不是来回发送到数据中心或云,而是本地处理

. 以零售商为例,他们现收银台使用边缘处理进行视频监控,这不仅是为了最大限度地减少产品损失,而且是为了结账时针对其他客户服务问题

当网络边缘的物联网继续取得进展时,资源限制给这些设备带来了巨大的挑战

物联网(IoT)设备如今受益于边缘较低的网络延迟和设备上更高的数据智能。这可以实现多种任务,从自动驾驶到实时视频流,再到设备的预防性维护。边缘处理绕过了集中式计算的时间延迟和数据安全挑战:数据不是来回发送到数据中心或云,而是本地处理

公司开始以五年前几乎无法想象的方式从边缘处理中获益。以零售商为例,他们现收银台使用边缘处理进行视频监控,这不仅是为了最大限度地减少产品损失,而且是为了结账时针对其他客户服务问题

因此,边缘处理正形成自己的体系结构,补充了需要实时处理和较低延迟的云体系结构。分析人士预测,这种情况将持续下去。虽然企业生成的数据中有10%是传统的集中式数据中心或云之外创建和处理的,但是 2025年,高德纳预测这一数字将达到75%

“已经踏上数字化商业之旅的组织已经意识到,需要一种更加分散的方法来满足数字化商业基础设施的需求,”said Gartner高级研究总监Santhosh Rao

“随着数据量和速度的增加,将所有这些信息流到云或数据中心进行处理的效率也会降低。”

边缘物联网有限的计算和电力资源

同时,边缘计算体系结构也受到各种计算和功耗的限制。边缘设备通常具有比数据中心和云资源更小的外形尺寸,并且可能位于难以访问的位置。这导致了权力和计算限制,限制了他们的效力

这是有问题的:边缘的数据密集型流程,如视频流、数据分析和自动驾驶,已经变得越来越突出,但这些任务也是需要边缘资源的数据大户

一个关于边缘计算、连接性和人工智能之间关系的会议上,专家们讨论了边缘物联网设备新发现的优缺点 嵌入式物联网世界 四月下旬

边缘物联网设备的紧凑性和较低的延迟会带来新的挑战。特别是,设备的远程性会对资源构成挑战

“这些传感器处于网络的最边缘,”科琳·约瑟夫森博士说。斯坦福大学电气工程专业候选人,嵌入式物联网世界边缘物联网设备会议演讲人。”工业湿度传感器……监测、街道污染监测、生态和农业监测——所有这些地方都网络的外围,因此很难联系到它们。”

受限的连接和远程访问与受限的电源配对

“如果你建筑物的墙壁深处有一个传感器,它的设计目的是发生泄漏时向你发出警报,你如何给它供电?约瑟夫森指出:“你如何将连接带到这些遥远的网络?”

阅读完整的文章>

 

英文译文:

While IoT at the edge of the network continues to make strides, resource constraints pose ample challenges to these devices.

Internet of Things (IoT) devices today benefit from lower network latency at the edge and greater data intelligence on device. This can enable a variety of tasks, from autonomous driving to real-time video streaming to preventative maintenance of equipment. Processing at the edge circumvents the time delays and data security challenges of centralized computing: Instead of sending data back and forth to a data center or a cloud, data is processed locally.

Companies are beginning to reap the benefits of edge processing in ways they barely imagined five years ago. Consider retailers, which now use edge processing for video surveillance at the register — not only to minimize product loss but also to target other customer services issues in checkout.

As a result, edge processing is coming into its own, complementing cloud architectures where tasks need real-time processing and lower latency. Analysts predict this will continue. While some 10% of enterprise-generated data is created and processed outside a traditional centralized data center or cloud, by 2025, Gartner predicts this figure will reach 75%.

“Organizations that have embarked on a digital business journey have realized that a more decentralized approach is required to address digital business infrastructure requirements,” said Santhosh Rao, senior research director at Gartner.

“As the volume and velocity of data increases, so too does the inefficiency of streaming all this information to a cloud or data center for processing.”

Constrained Compute and Power Resources for IoT at the Edge

At the same time, edge-computing architecture suffers from various compute and power constraints. Edge devices often have smaller form factors than data center and cloud resources and may be located in hard-to-access locations. That results in power and compute constraints that limit their efficacy.

That’s problematic: Data-intensive processes at the edge, such as video streaming, data analytics and autonomous driving, have become increasingly prominent, but these tasks are also data hogs that require resources at the edge.

Experts discussed the newfound pros and cons of IoT devices at the edge in a session on the relationship between edge computing, connectivity and AI at Embedded IoT World in late April.

The compactness and lower latency of IoT devices at the edge can bring new challenges. Particularly, devices’ remoteness can pose challenges for resources.

“These sensors are at the very edge of the network,” said Colleen Josephson, Ph.D. candidate, electrical engineering, Stanford University and speaker at a session on IoT devices at the edge during Embedded IoT World. “Industrial moisture sensors … monitoring, pollution monitoring on streets, ecological and agricultural monitoring — all these places are at the outer reaches of the network, so they are going to be hard to contact.”

Constrained connectivity and remote access are paired with constrained power.

“If you have a sensor deep in the walls of a building and it’s designed to alert you if there is a leak, how do you power it? How do you bring connectivity to these far reaches of the network?” Josephson noted.

 

Share this Post:

相关资讯: