Intelligent Intrusion Detection Based on Federated Learning for Edge-Assisted Internet of Things

Table 7

Comparison of classification performance between the local model and the global collaboration model.

Model; B = 128, K = 10, C = 1, η = 1e − 2	Local model: E = 40; FL model: E = 1, R = 40		Local model: E = 120; FL model: E = 1, R = 120		Local model: E = 200; FL model: E = 1, R = 200
	Accuracy (%)	Precision (%)	Accuracy (%)	Precision (%)	Accuracy (%)	Precision (%)
Client model01	99.07	92.38	99.22	93.04	99.54	93.99
Client model02	98.97	91.95	99.09	91.90	99.53	92.30
Client model03	99.06	91.31	99.27	92.01	99.57	94.53
Client model04	99.08	91.67	99.15	91.58	99.53	91.12
Client model05	99.06	90.11	99.21	92.92	99.56	93.78
Client model06	98.96	91.64	99.23	92.19	99.56	91.03
Client model07	98.98	91.50	99.18	93.01	99.58	94.17
Client model08	98.84	90.89	99.11	90.70	99.36	88.59
Client model09	99.06	90.86	99.22	91.52	99.57	91.62
Client model10	99.03	91.42	99.19	92.20	99.57	91.27
FedACNN	99.12	92.74	99.45	93.06	99.63	94.56