IEEE Systems, Man and Cybernetics Magazine - January 2023 - 35

Future Challenges in Cognitive
Network Architecture Systems
Despite the recent efforts described previously, some
future challenges in cognitive network architecture systems
still need to be addressed.
The Generalization and Effectiveness
of Incentive Mechanisms
How to effectively design the incentive mechanism for idle
smart devices with diverse incentive requirements, so as to
obtain the best incentive effect, will be a big challenge. This
will be related to whether the cognitive network architecture
system can gather enough idle smart devices, which
directly affects the ability of the system to construct cognitive
network services.
The Timeliness of Cognitive Network Services
As the data collection points are randomly distributed, the
processing of data needs to be completed by edge nodes
with certain computing power, even if it does not completely
rely on the cloud center. Therefore, how to design the
cognitive network architecture system by balancing the
factors, such as data collection delay, data transmission
delay, and data processing delay, will be a big challenge.
This will directly affect the timeliness of cognitive network
service production.
The Trustworthiness of Cognitive Network Services
How to collect trusted data and aggregate them to trusted
devices at a controllable cost in a heterogeneous and
untrusted environment will be a big challenge. This has a
direct impact on whether the built cognitive network services
can be trusted.
In addition, the deeper research on cognitive network
self-organization and resource dynamic intelligence optimization
is also one of the challenging issues in the future.
This will be explored by applying deep reinforcement
learning or policy-based reinforcement learning approaches,
such as policy gradient and actor-critic policy, to it.
Conclusion
Although cognitive networks are rich in spectrum resources,
communication devices, and computing resources, current
cognitive network resources are not fully utilized and, thus,
performance is seriously affected due to the fact that there is
no effective mechanism for deep sharing among primary
users, secondary users, resource owners, and resource users.
Aiming to address this issue, we discuss an intelligent selforganization
approach to build cognitive networks through
the data-driven cognitive network architecture system. In
such an architecture, spectrum resources, communication
devices, and computing resources are organized effectively,
together forming a self-organizing cognitive network and,
thus, network performance can be effectively improved without
additional costs. We further analyze the performance of
the architecture system by the simulation results, which outperform
the conventional scheme in terms of throughput,
delay, energy efficiency, and continuous service capacity.
Finally, we discussed some future challenges for cognitive
network architecture systems and their possible solutions.
Monitoring
Device
Waste Gas Monitoring
Macro Base
Station
Traffic Monitoring
Sewage Monitoring
Power Monitoring
Figure 5. The applications of the proposed scheme.
January 2023 IEEE SYSTEMS, MAN, & CYBERNETICS MAGAZINE 35
IEEE Systems, Man and Cybernetics Magazine - January 2023

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