Relevant node discovery and selection approach for the Internet of Things based on neural networks and ant colony optimization

The Internet of Things (IoT) brings opportunities to create new services and change how services are sold and consumed. The IoT is overpopulated by a large number of networks, millions of objects and a huge number of services and interactions. Despite this, the nature of IoT networks, such as the heterogeneity of resources, the dynamic topology, and the large number of similar services, makes service discovery a complex task in terms of accuracy and the time required. Furthermore, the discovery task can offer a set of providers for a given request, so selection of the most relevant provider node must take into account the available resources, such as the power energy and the period of time. In this paper, to overcome these limitations, we propose an approach for service discovery and selection in the IoT. The discovery phase is performed by an edge server using a neural network. The selection phase is performed by nodes to select the most adequate node from the set of relevant nodes using Ant Colony Optimization (ACO). The experimental results show high performance in term of accuracy (96.5%) and a longer network lifetime for the discovery and selection phases respectively, as well as a short period of time for both phases.