A multi-agent based optimised server selection scheme for SOC in pervasive environment

B Choudhury, P Dey, A Dutta, S Choudhury - Advances in Practical …, 2014 - Springer
Advances in Practical Applications of Heterogeneous Multi-Agent Systems. The …, 2014Springer
Recently immense research efforts has been observed towards development of multi-agent
based middleware for enabling service oriented computing (SOC) in mobile ad hoc and
pervasive environment. The fundamental issues addressed are service discovery, service
composition, service replication, service revocation and proper assignment of service
request to servers. But, very few proposal considered the total social welfare (sum total of
individual utilities) maximisation and self-organised emergence of such systems. In this …
Abstract
Recently immense research efforts has been observed towards development of multi-agent based middleware for enabling service oriented computing (SOC) in mobile ad hoc and pervasive environment. The fundamental issues addressed are service discovery, service composition, service replication, service revocation and proper assignment of service request to servers. But, very few proposal considered the total social welfare (sum total of individual utilities) maximisation and self-organised emergence of such systems. In this paper we propose a multi-agent based server selection scheme that ensures maximum social welfare and self organised system emergence by meeting the Quality of Service (QoS) requirement of all service consumers. To achieve this, we formalise it as a multi-agent based distributed constraint optimization problem (DCOP) and solve it using max-sum algorithm. The proposed max-sum algorithm offers significant saving in system wide resource usage and at the same time achieves lower service drop rate and slightly better service completion time. However, the message passing overhead for the max-sum algorithm is higher that limits its practical applicability. To overcome this, we propose a heuristic solution which gives similar gain as offered by max-sum algorithm with substantially lower message passing overhead.
Springer
Showing the best result for this search. See all results