International Journal of Engineering and Technology

Increase in digitalization has led to an increase in the data available for digital processing. Such data are termed as rich data, as they depict direct real time data. Difficulty arises when trying to process such data due to their inconsistent nature. This paper presents a density based clustering technique that can be used to identify arbitrary shaped clusters in data. The advantage of this approach is that it requires no external input to identify the range threshold. Particle Swarm Optimization is used as the selection technique to identify nodes belonging to a cluster. A multi-start variant of the Particle Swarm Optimization technique is used, which parallelizes the entire clustering process making it faster and more efficient. Experiments were conducted and it was identified that the current approach exhibits faster clustering process with better efficiency.