Optimized Placement of Scalable IoT Services in Edge Computing

Edge computing is a promising concept to enable the Internet of Things (IoT) vision, especially for supporting time-sensitive applications. A challenge in this topic is the service placement problem, which concerns the decision of where to place multiple applications according to their Quality of Service (QoS) requirements on the one hand and the computing resource availability on the other hand. In this paper, we jointly investigate the load distribution and placement of scalable IoT services, both vertically and horizontally, to minimize the potential violation of their QoS requirements due to the limitations of edge computing resources. We formulate the problem as an integer nonlinear programming. In order to handle the computational complexity, we propose two approaches, one through linearization techniques and another based on genetic algorithm. Experiment results show that our linearization proposal has low levels of violation in a small-scale scenario, and both proposals outperform other methods in a large network.