Time-optimized sequential decision making for service management in smart city environments

Edge Computing is a new computing paradigm that aims to enhance the Quality of Service (QoS) of applications running close to end users. However, edge nodes can only host a subset of all the available services and collected data due to their limited storage and processing capacity. As a result, the management of edge nodes faces multiple challenges. One significant challenge is the management of the services present at the edge nodes especially when the demand for them may change over time. The execution of services is requested by incoming tasks, however, services may be absent on an edge node, which is not so rare in real edge environments, e.g., in a Smart Cities setting. Therefore, edge nodes should deal with the timely and wisely decision on whether to perform a service replication (pull-action) or tasks offloading (push-action) to peer nodes when the requested services are not locally present. In this paper, we address this decision-making challenge by introducing an intelligent mechanism formulated upon the principles of Optimal Stopping Theory and applying our time-optimized scheme in different scenarios of services management. A performance evaluation that includes two different models and a comparative assessment that includes one model are provided found in the respective literature to expose the behavior and the advantages of our approach which is the optimal stopping theory (OST). Our methodology (OST) showcases the achieved optimized decisions given specific objective functions over services demand as demonstrated by our experimental results.