Search

The meta-heuristic algorithms require a lot of fitness calculations to get good enough solutions, which constitutes an obstacle to solving computationally complex practical problems. Recently, it has been found that surrogate-assisted meta-heuristic algorithms show potential in solving expensive complex optimization problems. This paper proposes an efficient surrogate-assisted Taguchi salp swarm algorithm (SATSSA) to solve expensive complex optimization problems. The SATSSA uses a combination of the local surrogate-assisted model (LSAM), global surrogate-assisted model (GSAM), and k-means clustering surrogate-assisted model (KCSAM) to fit the fitness function. To enhance the prediction ability of the model, an improved salp swarm algorithm with the Taguchi method (TSSA) is proposed to update and predict the model. GSAM is mainly used to capture the entire landscape of the search space. KCSAM is designed to capture part of the search space to improve the exploration capability of the algorithm. LSAM is used to capture the contours around the optimal individual. The proposed SATSSA is compared with other four excellent algorithms on 30D, 50D, and 100D benchmark functions. In addition, SATSSA is also applied to intrusion detection. Simulation results show that SATSSA is effective in improving detection rate and reducing false alarm rate. [ABSTRACT FROM AUTHOR]

Internet of Things (IoT) as an emerging technology is widely used in various applications such as remote healthcare, smart environment, and intelligent transportation systems. It is necessary to address users' concerns about cost, ease of use, privacy, and comprehensive security to grow the popularity of this technology. Intrusion Detection System (IDS) plays an indispensable role in security and preventing unauthorized users to access authorized network resources through analyzing network patterns. Several techniques such as metaheuristic algorithms, machine learning, fuzzy logic, and artificial intelligence algorithms can be applied to increase the accuracy of IDS, feature selection, and network patterns classification. In this paper, a hybrid intrusion detection system based on Support Vector Machine (SVM) and Grey Wolf Optimization (GWO) is presented that utilizes the advantages of these algorithms. In the proposed approach, the support vector machine has been used to train and differentiate anomaly records from normal records and grey wolf optimization has been used to find the kernel function, feature selection, and adjust optimal parameters for the SVM in order to improve the classification. The conducted simulations prove that the proposed approach outperforms in terms of detection accuracy, precision, recall, and F-score on both NSL-KDD and TON_IoT datasets. [ABSTRACT FROM AUTHOR]

The design of underwater wireless sensor networks (UWSNs) faces many challenges, including power consumption, storage, battery life, and transmission bandwidth. UWSNs usually either use node clustering or multi-hop routing as their energy-efficient optimization algorithms. The cluster optimization technique will organize the sensor nodes into a cluster network, with each cluster led by a cluster head (CH). In contrast, the multi-hop optimization algorithm will create a multi-hop network by sending data to the base station (BS) while switching between different sensor nodes. However, the overburdens of CH nodes impact the performance of the cluster optimization method, whereas the overburdens of nodes close to the BS impact the performance of the multi-hop optimization algorithm. Therefore, clustering and routing procedures can be considered as a simultaneous NP-hard problem that metaheuristic algorithms can address. With this motivation, this paper proposes an energy-efficient clustering and multi-hop routing protocol using the metaheuristic-based algorithm to increase energy efficiency in UWSNs and lengthen the network life. However, the existing metaheuristic-based methods use two separate algorithms for clustering and multi-hop routing, increasing computational complexity, different initialization, and difficulty in hyperparameters' tuning. In order to address the mentioned shortcomings, this paper proposes a novel hierarchical structure called hierarchical chimp optimization (HChOA) for both clustering and multi-hop routing processes. The proposed HChOA is validated using various metrics after being simulated using an extended set of experiments. Results are compared to those from LEACH, TEEN, MPSO, PSO, and IPSO-GWO to validate the impact of the HChOA. According to the findings, the HChOA performed better than other lifespan and energy usage benchmarks. [ABSTRACT FROM AUTHOR]