Skip to main content
. 2026 Mar 11;26(6):1780. doi: 10.3390/s26061780
Symbol Description
D Training dataset
xi Feature vector of sample i
yi Class label of sample i
C Number of attack categories
T Teacher network
S Student network
θT,θS Parameters of teacher/student networks
LKD Knowledge distillation loss
LCE Cross-entropy loss
LKL Kullback–Leibler divergence
T Temperature parameter for distillation
α Loss balancing coefficient
qb Quantization bit-width
τt Threat level at time t
β Exponential smoothing parameter
W Window size for threat estimation
EIDS Energy consumption of IDS module
EMAC Energy per multiply–accumulate operation
Emem Memory access energy per neuron
γnt Carbon intensity at node n, time t
ρnt Renewable energy fraction
xj,n,t Scheduling decision variable
Ej Energy requirement of job j
Pnmax Maximum power capacity of node n
Dj Deadline of job j
ωj Security priority weight
K Total number of participating nodes
G Number of node groups
Fkθ Local objective function
Fθ Global objective function
Acronym Definition
IDS Intrusion Detection System
EEIDM Energy-Efficient Intrusion Detection Module
LCE Lightweight Cryptographic Engine
HFLC Hierarchical Federated Learning Coordinator
CASE Carbon-Aware Scheduling Engine
KD Knowledge Distillation
FL Federated Learning
ASCON Authenticated Encryption Standard
ECDH Elliptic Curve Diffie-Hellman
AUC Area Under ROC Curve
ROC Receiver Operating Characteristic