IEEE Systems, Man and Cybernetics Magazine - January 2023 - 31

for the use of cognitive frequency and idle smart devices, a
more globally optimized scheduling can be achieved,
which is crucial to maintain the stability of the cognitive
network topology.
A Spectrum and Device Cognitive Engine as
Well as a Networking Cognitive Engine
To enhance the stability of cognitive networks, it is necessary
to perform the unified allocation and treatment of
the available spectrum resources and idle smart devices,
which leads to a comprehensive arrangement of the available
network resources. Based on the proposed architecture,
to improve the stability of future cognitive networks
and, thus, meet the requirements for sustainable ultrahigh
network capacity and massive online connections, we
present the spectrum and device cognitive engine as well
as the networking cognitive engine, as shown in Figure 2.
The spectrum and device cognitive engine can achieve
spectrum cognition (e.g., the dynamic spectrum utilization
opportunities), use cognitive techniques to uncover
spectrum usage patterns, and make predictions and guide
more effective spectrum reusing. Also, it can achieve the
cognition of device rental intention to make full use of
idle smart devices.
The networking cognitive engine can achieve the cognition
of secondary-user-based big data and then realize
intelligence associated with the secondary user requirements
(e.g., application-specific cognitive networks).
Through the spectrum and device cognitive engine as well
as the networking cognitive engine, cognitive spectrum
resources and idle smart devices can be well matched
with networking requirements, which can ensure the stability
of the cognitive network topology, achieve the network
densification goal, and lay a foundation for
obtaining low latency and ultrareliable connections. In
this article, to allocate spectrum resources and idle smart
devices more rationally for future cognitive network construction,
we argue that a broker should rent a hierarchical
cognitive computing facility. That is, the network edge
computing facilities mainly provide services for data preprocessing
and preliminary analysis, while the computing
facilities on the remote cloud are responsible for intensively
and deeply processing data by utilizing the global
information. The two cognitive engines can be modeled
by using machine learning technology, which can be
trained and deployed in an end-edge-cloud collaborative
manner. Due to space constraints, the details are not listed
in this article.
Cognitive Network Products
Building Cognitive
Networks for Sale
Building
Obtaining Available
Spectrum and
Communication Devices
Mining Spectrum and Device Rule Refining Specific Requirements
Renting Computing
Facilities
Available Spectrum
and Devices Rule Library
Matching
Network Service Requirements Set
Designing Machine
Learning Solutions
Spectrum and Device Cognitive Engine
Networking Cognitive Engine
Providing Spectrum and Devices
Providing Original Requirements
Idle Devices
Purchasing
Spectrum and
Idle Devices
Licensed
Spectrum Owners
Figure 2. A data-driven construction architecture for future cognitive wireless networks.
January 2023 IEEE SYSTEMS, MAN, & CYBERNETICS MAGAZINE 31
Cognitive Network Demanders
Data-Acquisition Layer
Information Mining Layer
Network Service Layer
IEEE Systems, Man and Cybernetics Magazine - January 2023

Table of Contents for the Digital Edition of IEEE Systems, Man and Cybernetics Magazine - January 2023
