example, in some given scenarios, the scale of the network is limited that means the
elements of the network are small, typically less than 100, so the evaluation of the
network is simple under the ability of computers and methods. In large-scale complex
networks, both academic and industry have launched some related researches to get the
precise estimation of the network. In [6], the authors provide the evaluation in WLAN
and 3G hybrid networks, to help select proper network and transmit data. In [7], the
virtualization of network is given, and the evaluation in core network is performed
under the digital modeling. In [8], the topology of the network is modeled complex
considering the wireless sensor scenario, a large number of wireless nodes with low
network requirements are modeled, but the scale and the complexity are not satisfied. In
[9], modeling of data traffic is given in hybrid network is given, and performance
evaluation based on mathematical traffic modeling is also presented. In [10], the
evaluation of multichannel using back-off method is given, but the scale is not suitable
for complex network evaluation. Then, in [11, 12], the authors provide some attempts
in evaluation of large-scale wireless networks, but the statistics methods seem complex.
In [13], SDN and RPL in wireless networks are evaluated based on the statistical
method, in [14], to tackle the problem of scale, the author provides the machine
learning way to solve the evaluation, and similarly, authors in [15–17] are trying to
evaluate the complex network using deep learning-based intelligent methods.
In this paper, to solve the problem of evaluation in large-scale networks, we propose the artificial intelligent (AI)-based method to simplify the evaluation using
parameter detection and ranking. The rest of the paper is organized as follows. In
Sect. 2, we provide the system model of complex wireless networks and the difficulties
in evaluation; in Sect. 3, we provide the methods of evaluation using deep learning, and
then, in Sect. 4, we provide the simulation and analysis. Conclusions are drawn as
summary.
2 System Model
The development of the fifth- and the sixth-generation wireless networks provides
significant improvement of network scale and the number of mobile terminals. The
evaluation of complex networks is quite difficult because the related parameters are
quite huge to compute.
In Fig. 1, one example of complex wireless networks is given. As is described
above, the scale of the network is one possible difficulty to evaluate the network
performance. It is clear in this figure that the topology is scaled into three main parts. In
cloud part, there are mainly the server and signal processing units, which will be the
content provider and the cloud signal processing. In this part, many types of data traffic
will be gathered, due to the requests from terminal users, signal processing in cloud
side will decide the optimal resource allocation and transmission paths to reduce the
waste of energy and system resource. In edge side, wireless transmission equipment
such as eNodeBs, wireless access points (WAP), wireless remote radio head (RRH) is
working in this part. In 5G and related advanced wireless networks, number of
equipment in edge side has become huge because of the development of MIMO,
network slicing, etc. In this part, main signal processing part has been transferred to
AI-Assisted Complex Wireless Network Evaluation Using Dynamic …
547
elements of the network are small, typically less than 100, so the evaluation of the
network is simple under the ability of computers and methods. In large-scale complex
networks, both academic and industry have launched some related researches to get the
precise estimation of the network. In [6], the authors provide the evaluation in WLAN
and 3G hybrid networks, to help select proper network and transmit data. In [7], the
virtualization of network is given, and the evaluation in core network is performed
under the digital modeling. In [8], the topology of the network is modeled complex
considering the wireless sensor scenario, a large number of wireless nodes with low
network requirements are modeled, but the scale and the complexity are not satisfied. In
[9], modeling of data traffic is given in hybrid network is given, and performance
evaluation based on mathematical traffic modeling is also presented. In [10], the
evaluation of multichannel using back-off method is given, but the scale is not suitable
for complex network evaluation. Then, in [11, 12], the authors provide some attempts
in evaluation of large-scale wireless networks, but the statistics methods seem complex.
In [13], SDN and RPL in wireless networks are evaluated based on the statistical
method, in [14], to tackle the problem of scale, the author provides the machine
learning way to solve the evaluation, and similarly, authors in [15–17] are trying to
evaluate the complex network using deep learning-based intelligent methods.
In this paper, to solve the problem of evaluation in large-scale networks, we propose the artificial intelligent (AI)-based method to simplify the evaluation using
parameter detection and ranking. The rest of the paper is organized as follows. In
Sect. 2, we provide the system model of complex wireless networks and the difficulties
in evaluation; in Sect. 3, we provide the methods of evaluation using deep learning, and
then, in Sect. 4, we provide the simulation and analysis. Conclusions are drawn as
summary.
2 System Model
The development of the fifth- and the sixth-generation wireless networks provides
significant improvement of network scale and the number of mobile terminals. The
evaluation of complex networks is quite difficult because the related parameters are
quite huge to compute.
In Fig. 1, one example of complex wireless networks is given. As is described
above, the scale of the network is one possible difficulty to evaluate the network
performance. It is clear in this figure that the topology is scaled into three main parts. In
cloud part, there are mainly the server and signal processing units, which will be the
content provider and the cloud signal processing. In this part, many types of data traffic
will be gathered, due to the requests from terminal users, signal processing in cloud
side will decide the optimal resource allocation and transmission paths to reduce the
waste of energy and system resource. In edge side, wireless transmission equipment
such as eNodeBs, wireless access points (WAP), wireless remote radio head (RRH) is
working in this part. In 5G and related advanced wireless networks, number of
equipment in edge side has become huge because of the development of MIMO,
network slicing, etc. In this part, main signal processing part has been transferred to
AI-Assisted Complex Wireless Network Evaluation Using Dynamic …
547