Edge Intelligence: The Central Cloud is Dead – Long Live the Edge Cloud!
Need
Industry sector
Manufacturing Buildings & Facilities Energy & Utilities Consumer & Home
Mobility 55 10
10
55
Ultra Low latency (<10ms) 95 85
5
15
Autonomy 95 100
100
50
Security 100 100
100
25
Local Network Bandwith 100 90
10
35
WAN Network Bandwith 35 55
10
55
Peer-to-Peer
Communication 80 85
50
90
Prioritization 100 15
90
10
Self-organization,
discovery 60 20
40
65
100
85
45
Artificial
Intelligence/
100
Machine Learning
Table 1: High level needs for edge intelligence by industry sector
authentication
and
authorization
running on edge computing units are
envisioned to enable fast adaptability of
the systems.
Prioritization: refers to prioritizing the
communication information depending
on the type of included data (data
gathering, alarms).
Self-organization, discovery: discovery of
the capabilities of the devices and
services and their role in the
infrastructure, so that operations can be
handed over from humans to intelligent
software.
analyzes data on edge computing units can
deliver actions and awareness with minimal
latency. A broad view of industry needs on
different capabilities of the edge intelligence
is summarized in Table 1 using a heat map to
suggest importance for a specific industry.
The scale ranges from 0 to 100, with 0 being
unnecessary and 100 value in case the
feature is vitally important for the industry.
The assignment of the values is done based
on a discussion between the authors of the
article, having as background the previous
projects.
Hardware Evolution
Artificial intelligence/machine learning: may
require and can gain from algorithms that
automate decisions or alarms without
human intervention. Adaptive software that
IIC Journal of Innovation
The trend within data centers is to shift to
smaller, more agile (i.e. movable) data
centers deployed towards the edge. This
includes, for example, “edge caching”
- 19 -