IEEE Computational Intelligence Magazine - November 2019 - 76

Operative Domains
Technology Providers
Open Source
and Proprietary
Software
Solutions

Infrastructure
From Cloud
Providers and
Data Centers

Interoperability Space

Devices and Execution

Intelligence Functions

Dominant Players and
Agreements Define
the Intelligence
Interoperability

Proprietary Specialized
Devices
Hardware

Tailor-Made
Domain Specific
Solutions

Main Focus
On Data
Ingestion

Mostly
Proprietary
Data Sources

Client-Server
Type of Applications

Machine
Intelligence
Ecosystem

Today's Dominant Business Models
Intelligence
Orchestration
Function

Intelligence
Providers

Value Chain

Proprietary and Open Models
From Generic or Specialized
Modules

Standardized Intelligence
Modelling and Exposure
to Applications

Local, Remote, Public and
Private Infrastructure
Proprietary, Third Party
or Local Data Sources

Intelligent
Devices

Intelligence
Market

Harmonized Intelligence
Life Cycle Management
and Provisioning Discovery

Open Intelligence
Services Platform

Intelligent
Services

Intelligent
Applications

Mostly Composition
of Generic Solutions

Services or Support to Peer-to-Peer, Multihoming,
Stand-Alone and Clientany Hardware Type
Server Applications

Machine Intelligence Ecosystem Business Enablers
Service Creation

Enabling Technologies
FIGURE 9 Machine Intelligence Distribution value generation enablers.

VM/container. In this last option, the
intelligence layer implementation
resembles a series of containers that
each has its own API that is accessible
by the application layer. This system
would replace a more monolithic entity
that provides a general intelligence service to higher layers. The pre-configured VM/container option would also
present possibilities to procure independent and protected memory areas and
restricted accessibility to the configuration and control interfaces.
2) Semantic Interoperability
An additional aspect to consider is the
meaning of the data used for intelligence. The actual interpretation of a data
set may vary between systems, which
can lead to confusing or incorrect inferences. Semantic interoperability between
the data models used by the data sources
and the understanding of the data by the
intelligence models is necessary and
important. Current efforts in the IoT
space, such as Light Weight Machine to

76

Machine6 (LWM2M) or World Wide
Web Consortium (W3C) Web of Things7,
are working to provide frameworks,
resources, and tools that potentially address
not only the problem of understanding
the nature of the data, but also to discover devices' capabilities. These initiatives
describe possible interactions with the
devices and how to understand their
operational requirements and possible
outputs. For intelligent devices and platforms, this knowledge also impacts the
capabilities descriptions. A device should
be able to unambiguously inform what
intelligent actions and services it can
provide, and its capabilities to support
intelligent algorithms [29]. Additionally,
semantic is needed to describe intelligent services to applications, and what
are the required input and expected output that enables intelligence discovery.
The goal is to provide a semantic description that should be understandable and
6
7

http://www.openmobilealliance.org
https://www.w3.org/TR/wot-architecture/

IEEE COMPUTATIONAL INTELLIGENCE MAGAZINE | NOVEMBER 2019

operational for automatic intelligence configuration without human intervention.
V. Conclusion

The spread of intelligence in all types of
devices, from the consumer to industrial
domains, requires specialized solutions
that nowadays are provided by proprietary means (with very little interoperation) and supported by centralized
systems. This tendency does not seem
sustainable, especially when the number
of devices and their capabilities are dramatically increasing. On the grounds of
the presented limitations, we propose
introducing an intelligence layer to the
devices stack that allows multiple applications to benefit from the intelligence
setup provided by intelligence providers.
The introduction of an intelligence layer
promotes an intelligence distribution
ecosystem with an MI distribution framework operating on top of it. This ecosystem would enable the development of
(continued on page 92)
http://www.openmobilealliance.org https://www.w3.org/TR/wot-architecture/
IEEE Computational Intelligence Magazine - November 2019

Table of Contents for the Digital Edition of IEEE Computational Intelligence Magazine - November 2019
