|
| Deep learning in wireless communication | Deep learning in wireless sensor network |
|
In fault detection
Finding fault in wireless communication using deep learning is done by the method of deep neural networks, which can address the intricacy of various networks. It can achieve it in less charge where the examination or experimental charge is low [32]. | In fault detection
The naive Bayes classifier and convolution neural network (CNN) increase the conjunction concert and determine the default nodes where the real-world dataset is used. It is also good in the identification of faults [33]. |
In channel state data
It needs an excessive amount of computational density, which makes it inconvenient to employ a novel technique. This can be sorted using the learning techniques of CNN, LSTM, RNN, etc., which can yield a constant estimation result [34]. | In channel state data
The data that received signals from the channel are complex in the handling of both key management and structural dependency. So, here it takes the method of LSTM, which produces the result accurately within time [35]. |
In quality prediction
It shows a hike in results of routing algorithms when it neglects unwanted links. However, quality prediction is a complicated method due to its variation in quantities of wireless infrastructure. To deal with this issue, the deep learning-based variants of RNN, LSTM, and GRU were resulted in upgrading in performance accuracy [36]. | In quality prediction
To have different system networking applications, such as booking and improved real-time video over 4G LTE networks, as well as digit rate transformation for improved execution in Wi-Fi networks. A succession deep learning model based on an encoder-decoder setup that is suitable for predicting future variations in distant signal strength from previous data analysis [37]. |
The location of the communication network
Because of high mobility, it accommodates heterogeneous necessities. Notwithstanding, the development of UAVs forces an interesting test for precise shaft arrangement between the UAV and the ground base station (BS). So, the LSTM-RNN technique is built for the UAV location issue [38]. | The location of the sensor network
The indoor restriction has received wide consideration as of late because of the expected utilization of wide scope of canny administrations. It is a profound learning-based methodology for indoor limitation using transmission channel quality measurements, including received signal strength and channel state [39]. |
Wireless communication security
In a communication situation, an attacker attempts to decide the balance plan of the blocked sign. It is used to limit the precision of the attacker, while ensuring that the expected recipient can in any case recuperate the basic message with the most elevated dependability. This is accomplished by bothering channel input images at the encoder, similar to antagonistic assaults against classifiers in machine learning [40]. | Wireless sensor network security
The wireless sensor switches and entryways are associated with the conveyed hubs to help some constant applications. Because of open access, the security issue emerges in WSN. WSNs are profoundly vulnerable to task assaults as it does not have the synchronization between hubs during information routing. A new lightweight DoS discovery conspires a deep learning-based defense mechanism (DLDM), which has proposed to recognize and detach the assaults in the data forwarding phase (DFP), and depicts the new calculation for the fruitful recognition of DoS assaults, such as weariness, sticking, homing, and flooding [41]. |
|
