Computational Intelligence (CI) has been addressed as a great challenge in recent years, particularly the aspects of routing, task scheduling, and other high-complexity issues. Especially for the Contact Plan Design (CPD) that schedules contacts in dynamic and resource-constrained networks, a suitable CI algorithm can be exchanged for a high-quality Contact Plan (CP) with the appropriate computational overhead. Previous works on CPD mainly focused on the optimization of satellite network connectivity, but most of them ignored network topology characteristics. In this paper, we study the CPD issue in the spatial node based Internet of Things (IoT), which enables the spatial nodes to deliver data cooperatively via intelligent networking. Specifically, we first introduce a Multi-Layer Space Communication Network (MLSCN) model consisting of satellites, High Altitude Platforms (HAPs), Unmanned Aerial Vehicles (UAVs), and ground stations, on which a Time-Evolving Graph (TEG) is used to illustrate the CPD process. Then, according to the characteristics of each layer in the MLSCN, we design the corresponding CPs for the inter-layer contacts and intra-layer contacts. After that, a CI algorithm named as Multidirectional Particle Swarm Optimization (MPSO) is proposed for inter-layer CPD, which utilizes a grid-based initialization strategy to expand the diversity of individuals, in which a quaternary search method and quaternary optimization are introduced to improve efficiency of particle swarms in iterations and to ensure the continuous search ability, respectively. Furthermore, an optimized scheme is implemented for the intra-layer CPD to reduce congestion and improve transmission efficiency. Simulation results show that the proposed CPD scheme can realize massive data transmission with high efficiency in the multi-layer spatial node-based IoT.
Keywords: Internet of Things; computational intelligence; contact plan design; delivery time; multidirectional particle swarm optimization.