Connectivity Aware and Energy Efficient Self-Organizing Distributed IoT Topology Control

TL;DR

This paper addresses topology control for distributed IoT networks under energy constraints by introducing a physics-inspired Hamiltonian framework. The network is modeled as a spatial graph and a cost function $H$ combines topology and communication terms, with Monte Carlo dynamics guiding the system toward high connectivity while minimizing energy consumption. The approach is instantiated across centralized static, ad hoc, and mobile ad hoc (MANET) network settings, producing distributed, self-organizing behavior that achieves near-100% connectivity with reduced transmission ranges and enhanced robustness to node failures and mobility. The work demonstrates significant improvements in connectivity stability and energy efficiency, and it outlines clear avenues for extending to 3D networks, interference considerations, and learning-based parameter optimization.

Abstract

Internet of Things has pervaded every area of modern life. From a research and industry standpoint, there has been an increasing demand and desire in recent years to develop Internet of Things networks with distributed structure. Wireless communication under emergency circumstances is one of the important applications that distributed Internet of Things can have. In order for a network to be functional in this scenario, it must be developed without the aid of a pre-established or centralized structure and operated in a self-organized manner to accommodate the communication requirements of the time. Although the design and development of such networks can be highly advantageous, they frequently confront difficulties, the most significant of which is attaining and maintaining effective connectivity to have reliable communications despite the requirement to optimize energy usage. In this study, we present a model for self-organizing topology control for ad hoc-based Internet of Things networks that can address the aforementioned challenges. The model that will be presented employs the notion of the Hamiltonian function in classical mechanics and has two key objectives: regulating the network's topology and dynamics to enhance connectivity to a desirable level while requiring the least amount of energy possible. The results of the simulation indicate that the proposed model satisfactorily fulfills the goals of the problem.

Figures (15)

• Figure 1: Euclidean distance is the connection metric employed here. In this network, a direct connection or link is created when two arbitrary nodes are within the transmission range of each other.
• Figure 2: (a) represents the initial spatial network constructed before any modifications at step 0. (b) represents the network topology following the execution of 15000 Monte Carlo steps with basic algorithm. (c) is the same network after execution of 15000 Monte Carlo steps with topology control algorithm. These snapshots are from a centralized static network simulation with the control parameters $\rho = 0.05, T = 100$, and $\gamma = 1$.
• Figure 3: Variation diagrams for (a) cost (b) connectivity (c) average scaled ranges$^2$ (d) average Tau over Monte Carlo steps is demonstrated for a centralized static network with the control parameters $\rho = 0.05, T = 100$, and $\gamma = 1$. These diagrams represent the average of ten statistical ensembles of the intended network.
• Figure 4: Variation diagrams for (a) cost (b) connectivity (c) average scaled ranges$^2$ (d) average Tau over Monte Carlo steps is demonstrated for a centralized static network facing node failures. The network is created with the control parameters $\rho = 0.05, T = 100$, and $\gamma = 1$. The diagram's colors represent the different node percentages of failed nodes. These diagrams represent the average of ten statistical ensembles of the intended network.
• Figure 5: Variation diagrams for (a) cost (b) connectivity (c) average scaled ranges$^2$ (d) average Tau over Monte Carlo steps is demonstrated for a 3D centralized static network with the control parameters $\rho = 0.05, T = 100$, and $\gamma = 1$. These diagrams represent the average of ten statistical ensembles of the intended network.