Energy Efficient Distributed Processing for IoT

In this paper, the entire IoT-fog-cloud architecture is modelled, the service placement problem is optimized through Mixed Integer Linear Programming (MILP) and the total power consumption is jointly minimized for processing and networking. Four aspects of IoT service placements are examined: 1) non-splittable services, 2) splittable services, 3) inter-service processing overhead for sub-service synchronization and 4) deployment of special-purpose cloud data centers (SP-DCs). The results showed that for a capacitated problem, service splitting introduces power consumption savings of up to 86% compared to 46% with non-splittable services in relation to processing in general-purpose data centers (GP-DCs). Moreover, it is observed that the inter sub-service processing overhead has a great influence on the total number of service splits. However much insignificant the ratio of the processing overhead, the results showed that this is not a trivial matter and hence much attention needs to paid to this area in order to make the best use of the resources that are available in the edge of the network. Moreover, the optimization results showed that, for very high demands, power savings of up to 50% could be achieved with SP-DCs compared to 30% with GP-DCs.