No abstract

Network security plays a critical role in our lives based on the threats and attacks to which we are exposed and that increase daily; these attacks result in the need to develop different protection methods and techniques. Network intrusion detection systems (NIDSs) are a way to detect several malicious network attacks. Many researchers have focused on developing NIDSs based on machine learning (ML) approaches to detect variants of attacks. ML approaches can automatically discover the essential variances between normal and abnormal data by analysing the features of a large dataset. Indeed, many features are extracted without discrimination, increasing the computational complexity. By applying a feature selection method, a subset of features is selected from the whole feature set with the aim of improving the performance of MLbased detection methods. The salp swarm algorithm (SSA) is a nature-based optimization algorithm that has demonstrated efficiency in minimizing processing challenges to perform optimization for feature selection problems. This research investigates the impact of the SSA on improving ML-based network anomaly detection using different ML classifiers, including XGBoost and Naïve Bayes algorithms. Experiments were conducted on standard datasets for comparison purposes; two datasets are used explicitly for network intrusion attacks: UNSW-NB15 and NSL-KDD. The experimental results show that the proposed method performs better in improving anomaly NIDSs in terms of the f-measure, recall, detection rate, and false alarm rate on both datasets. It outperforms the state-of-the-art techniques recently proposed in the literature.

Cloud computing has a great ability to store and manage remote access to services in a term of software as a service (SaaS). Recently, many organizations have moved to use outsourcing over the cloud to reduce the local resource burden. The stored services over the cloud are too scalable and complex, so an optimization method is more desirable to select appropriate services that satisfy the clients' request. To do so, the quality of service (QoS) parameters that associated with each service are the best resources for choosing and optimizing the appropriate services over the cloud. Therefore, the cloud service composition aims to select and integrate services over the cloud to satisfy the clients' request. In this work, a hybrid algorithm is introduced, which combines ant colony optimization (ACO) and genetic algorithm (GA) to efficiently compose the services over the cloud. The GA is used to tune the ACO's parameters automatically and the ACO adapts its performance based on the parameters tuning. The main contribution of this work is to help the ACO algorithm to avoid stagnation problem and enhance the performance of the ACO where this performance is affected by the value of the ACO's parameters. The experimental results on 15 different real datasets have shown the effectiveness of the proposed algorithm to search comparable solutions compared to five competitors.INDEX TERMS Cloud Services Composition (CSC); Ant Colony Optimization (ACO), Genetic Algorithm (GA).

No abstract