A survey of swarm-inspired metaheuristics in P2P systems: some theoretical considerations and hybrid forms

Abstract

The growing complexity of nowadays distributed systems influences the application of nature-inspired mechanisms as efficient problem-solving methods. They are important and inevitable for the optimisation and robustness of distributed systems, where autonomous agents interact without central control. Especially in the P2P systems, swarm-inspired techniques provide incentives and encourage cooperative behaviour between the peers. Many open problems in the P2P systems and cloud computing are characterised by huge and unforeseen dynamics, and number of unpredictable dependencies on participating components. Therefore, there is a demand on self-organising approaches. Swarm intelligence possesses distributive and autonomous properties, represents a self-organising biological system and swarm-inspired algorithms play an important role in the P2P systems and cloud computing. This survey paper presents an overview of swarm-inspired algorithms used in P2P systems and cloud computing, describes their underlying biological behaviours, their concept, working, and their main features. Further, the main intention of this paper is to give an overview of the theoretical background of such swarm-inspired metaheuristics in terms of asymptotical behaviour, convergence, etc. as well as a thorough overview of the existing hybrid forms (swarm-inspired metaheuristic with another swarm-inspired metaheuristic). In the scope of this survey paper, a new classification of swarm metaheuristics is proposed.