Comparing and Analyzing Applications of Intelligent Techniques in Cyberattack Detection

Table 2

Comparative study of metaheuristic algorithms.

Features	PSO	GA	ACO	AIS	ABC	BFA
Representation	Dimensional vector for position speed, the best state	Binary, real list of rules, permutation of elements	Undirected graph	Attribute str. (a real-valued vector), integer string, binary string symbolic string	D-dimensional vector (xi = 1, 2, …, D)	Represents i-th bacterium at jth chemotactic, k-th reproductive, and l-th dispersal step
Operators	initializer, update, and evaluator	Crossover, mutation, selection, inversion	Pheromone update and measure, trail evaporation	Immune operators cloning, hypermutation and selection based on elitism	Reproduction, replacement of bee, selection	Reproduction, chemotaxis, dispersion, elimination
Datasets for attack detection	KDDCUP99, DARPA98, NSL-KDD	KDDCUP99, DARPA98	KDDCUP99, DARPA98	KDDCUP99, DARPA98, LLDOS	KDDCUP99, DARPA98	KDDCUP99, DARPA98
Permittivity against DoS attack	High	Low	High	Low	Low	Low
Structure and dynamics	Discrete and network components and evolution or learning based	Discrete and network components and evolution based	Discrete components and evolution and learning based	Discrete and network components and evolution or learning based	Discrete and components and evolution based	Discrete and network components and evolution or learning based
Reliability	High	Low	High	High	Low	Low
Time-consuming	Time-consuming because of its complexity but the no. of repetitions is less	Low time-consuming because of its simplicity but the no. of repetitions is high	Time-consuming because of its complexity	Low time-consuming because of its simplicity	Low time-consuming because of its simplicity	Low time-consuming because of its simplicity
Robustness	High	Low as compared to PSO, but more than others	Lower than PSO and GA, more than others	Lower than ACO better	Lowest	Higher than ABS
Parameters	Number of particles, dimension of particles, range of particles, maximum number of iterations, inertia weight	Population size, max generation number, cross-over probability	Number of ants, iterations, pheromone evaporation rate, amount of reinforcement	Population size, no. of antibodies to be selected for hypermutation, number of antibodies to be replaced	No. of food sources which is equal to the no. of employed onlooker bees	The dimension of the search space, number of bacteria, number of steps chemotactic, no. of elimination and dispersal events, no. of reproduction steps, probability
Applications	Power system optimization problems, multimodel problems, multiobjective, dynamic, constrained, and combinatorial optimization problems, anomaly detection, sequential ordering problem, etc.	Pattern recognition, reactive power dispatch, sensor-based robot path planning, multiobjective vehicle routing problem, molecular modeling, web service selection, etc.	Continuous optimization and parallel processing implementations. Vehicle routing problem, graph coloring and set covering, agent-based dynamic scheduling, etc.	Computer security, anomaly detection, clustering/classification, numeric function optimization, virus detection, pattern recognition, etc.	Solving reliability redundancy allocation problem, training neural networks, XOR, decoder-encoder, and 3-bit parity benchmark problems, pattern classification, etc.	Application for harmonic estimation problem in power systems, the parameters of membership functions, and the weights of rules of a fuzzy rule set are estimated, etc.