Hybrid metaheuristic method for clustering in wireless sensor networks / Bryan Raj Peter Jabaraj
Wireless Sensor Networks (WSNs) are used widely in many applications to ease data access in large-scale and hard-to-reach areas. However, WSNs possess many limitations, such as limited energy, memory size and communication ranges. Energy is the biggest concern in WSNs, as these nodes are deployed ra...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Published: |
2023
|
| Subjects: | |
| Online Access: | http://studentsrepo.um.edu.my/15247/2/Bryan_Raj.pdf http://studentsrepo.um.edu.my/15247/1/Bryan_Raj.pdf http://studentsrepo.um.edu.my/15247/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Wireless Sensor Networks (WSNs) are used widely in many applications to ease data access in large-scale and hard-to-reach areas. However, WSNs possess many limitations, such as limited energy, memory size and communication ranges. Energy is the biggest concern in WSNs, as these nodes are deployed randomly in hard-to-reach sensing fields. So, the idea of replacing the battery is not a viable option. To alleviate the problem, clustering techniques were proposed in the early 2000s. However, it faced issues such as isolated node problems and energy hole problems because of the inefficiency in Cluster Head (CH) selection. As such, the existence of metaheuristic methods to optimally select the CH and forms clusters has opened up a research interest in proposing a metaheuristic method with balanced exploration and exploitation ability for efficient CH selection. As such, this thesis proposes a hybrid metaheuristic method that consists of Sperm Swarm Optimization (SSO) algorithm and Genetic Algorithm (GA), which is termed HSSOGA. To ensure the performance of the developed method in obtaining the optimized solution, the method is evaluated on 11 test benchmark functions named Sphere, SumSquare, Zakharov, Rosenbrock, Step, Ackley, Griewank, Rastrigin, Schwefel 2.26, Michalewicz and Egg Crate. The results obtained by the proposed HSSOGA in optimizing this function was promising as it ranked first in the majority of the test function compared to existing hybrid metaheuristic method such as HFPSO, HPSOGA, SAGA, PSOGWO, HSSOGSA and existing conventional methods termed SSO and GA. Then, the proposed HSSOGA is enhanced by adaptively tuning the crossover and mutation probability, as well as linearly reducing the velocity of the sperms to ensure the exploration and exploitation of the method are controlled based on network changes. The adaptive HSSOGA (aHSSOGA) is implemented in the WSN environment to mitigate the isolated node and energy hole problems. To assist the proposed method, the objective functions used to select optimal CH is refined by adding objectives such as CH’s maximum neighbour node and average isolated node probability. Moreover, two improvised clustering techniques are introduced to reduce the energy overhead cost from the re-clustering process. The performance of aHSSOGA is evaluated based on average residual energy, network lifetime, total re-clustering occurrence, total data delivery, network throughput and end-to-end delay metrics. The proposed aHSSOGA outperforms the state-of-the-art.
|
|---|