[Retracted] A Survey on Information-Centric Networking with Cloud Internet of Things and Artificial Intelligence

Table 2

Detailed AI/ML technique analysis.

Reference	ICN technologies	AI/ML technologies	Model framework	Simulation	Results
[26]	NDN	Deep Neural Network (DNN)	A feed-forward two-layer neural network with sigmoid units at the hidden layer and linear units at the output layer	ndnSIM	Accuracy of 99.23%
[17]	ICN	Machine Learning (ML)	Frequent pattern- (FP-) growth algorithm	Simulation environment, ICN server, edgeNode, and user	Reduces the time needed to access data and increases the efficiency of data collecting
[20]	ICN-V	Multiagent AI systems	Game theory analysis	Simulation mechanism of caching and sharing information in IoV based on evolutionary games	Helps information sharing in ICN-V
[27]	ICN	DNN	Deep learning as a black box technology and least recently used (LRF) or least frequently used (LFU)	No simulation	—
[19]	ICN	ML	Multilayer perceptron (MLP) for supervised learning, deep belief network- restricted Boltzmann machines (DBN-RBM) for unsupervised learning, and extended learning classifier system under reinforcement learning (RL)	Calculate root mean square error (RMSE) using MATLAB	XCS algorithm under RL methods outperforms the others
[23]	NDN	Artificial Neural Network (ANN)	Caching and replacement decisions by using model/parameter/inference	No simulation	—
[24]	ICN-V	Natural Language Processing (NLP) and ML	Semantic-based analysis	Testbed and EdgeCloudSim	Performs commendably in emergency situations
[25]	ICN-5G	ML	RL, Q-learning/Markov decision process	MATLAB	Based on their stretch profile over ICN, the technique finds the best routing path for delay-sensitive 5G-enabled TI. ISO strategy outperforms random routing by 33.69% and HARP by 33.33%
[29]	ICN-M	Neural network/ML	Image classification/convolutional network	ndnSim, Keras, and tensor for deep learning	Smart mechanism to monitor and counter COVID-19 pandemic spread in close indoor environments
[22]	ICN	ML	-means clustering	Icarus simulator (ICN caching simulator)	In comparison with the best-considered schemes, the proposed technique demonstrates a 15% increase in cache hit ratio, a 12% reduction in content retrieval delay, and a 28% reduction in average hop count