A Proactive Context-Aware Service Replication Scheme for Adhoc IoT Scenarios

We consider a smart-city IoT scenario where large crowd may gather temporarily rendering the existing infrastructure inadequate for service consumption. This necessitates a service replication framework over quasi-adhoc scenario using the available computing resources carried by the users such as smart-phones. Such framework can offer fog computing solution in addition to enabling consumption of plethora of new services available with the crowd. In this paper, we propose a service replication scheme that achieves improved service availability, service response-time and system-wide resource utilization compared to the existing ones. The scheme uses a dual-threshold-based proactive sensing mechanism to identify the services which are required to be replicated in immediate future and a multi-agent-based optimal task assignment scheme that enables batch-wise decision making. These mechanisms acting together reduces the service drop rate and improves the system-wide resource utilization. The service response time and the overhead involved in making assignment decisions, are markedly reduced by applying a strategy that combines the benefits of both physical and functional contexts together. An integrated model for analyzing the performance of various generic service replication schemes, is also developed in this paper.