An Approximation Mechanism for Elastic IoT Application Deployment

Abstract

Even though Internet of Things (IoT) applications are proliferating exponentially in recent years, there still exist several resource allocation problems in IoT application implementations with the traditional cloud computing. Because the number of IoT applications is increasing day by day with strict latency requirements. They become a burden to the cloud/datacenter platform in order to fulfill a huge number of real-time IoT services. As the emerging solution for latency requirements, the edges can bring processing power closer to datasource - the Thing in IoT. However, with the resource limitation at edges, the efficient resource allocation is a major concern to improve the performance of edge networks. In this work, we introduce the optimization model, named the Service-Oriented Resource Allocation (SORA) for IoT applications, which dynamically consolidates the system so as to reduce the system cost while improving the available resource at the edges. Unfortunately, SORA is unable to solve in polynomial time because it is NP-hard. Unlike the prior works that try to find solutions based on heuristic algorithms, we propose approximation algorithms to solve SORA with a near-optimal solution. Finally, we evaluate our model by providing several simulation cases, in which our proposed mechanisms show outstanding outcomes in terms of solving SORA and resource utilization.