Impact of Random Number Generation Methods Usage on Swarm Intelligence Algorithms for Energy Optimization in Wireless Sensor Networks

• Published in: 2024 IEEE 3rd Conference on Information Technology and Data Science (CITDS) pp. 1 - 8
• Main Authors: Filep, Levente, Gal, Zoltan
• Format: Conference Proceeding
• Language: English
• Published: IEEE 26.08.2024
• Subjects: Generators; Knowledge engineering; MATLAB; Measurement; Optimization; Particle swarm optimization; Problem-solving; Random number generation; Random Number Generators; Search problems; Swarm Intelligence; Wireless Network Sensors; Wireless sensor networks
• DOI: 10.1109/CITDS62610.2024.10791391

Swarm Intelligence (SI) is a complex, adaptive, and intelligent collective behavior observed in decentralized, self-organized systems. These behaviors arise from the collective, yet simple actions of individual agents forming the group. SI algorithms gained significant attention in various fields of science due to their optimization and problem-solving applications. A variety of algorithms have been proposed in the literature and applied to different optimization problems. As the actions of individuals are partly governed by seemingly random behavior, as well as the system usually being initialized at random, these algorithms rely on random number generators. These generators are pseudo-random in nature, an important aspect of experiment result repeatability. We often don't think about these generators affecting the algorithms' performance, especially after a large number of generated random numbers. However, as we will see in this paper, this is not always the case. This paper focuses not on the mathematical background of the RNG algorithms but on the effects of them on the SI algorithms' behavior conducted in MATLAB. We further focus on the performance of SI algorithms in WSN antenna placement problems, as well as classical benchmark landscapes, such as Rastrigin, and Rosenbrock.