Multi-Objective Ant Colony Optimization for Energy-Aware Routing and Load Balancing in Distributed IoT Meshes with Intermittent Connectivity

Abstract

The rapid growth of distributed Internet of Things meshes has intensified demands for routing strategies that can remain effective under intermittent connectivity while respecting strict energy budgets and avoiding persistent hot spots. Devices in such meshes are diverse in their capabilities and experience uneven traffic, heterogeneous link qualities, and duty-cycled radios that break end-to-end paths without warning. In this context, multi-objective methods can capture realistic trade-offs between energy usage, latency, reliability, and load distribution, but they must cope with time-varying graphs and partial information. Metaheuristics inspired by the collective behavior of biological systems are often robust to noise, and ant colony optimization is particularly appealing because it can couple local sampling with global reinforcement in a decentralized manner. This paper explores a multi-objective ant colony framework specialized for energy-aware routing and load balancing in distributed IoT meshes when links are intermittent. The study builds a linear-algebraic formulation of flows, constraints, and energy states, then designs a pheromone policy that integrates queue backpressure, residual energy, and contact likelihood. It analyzes stability properties under slowly varying channels, proposes lightweight estimators of path reliability and residual battery costs, and reports controlled simulations that examine the effects of evaporation rates, colony size, and scalarization schedules on convergence behavior. The approach highlights feasible compromises that limit peak node stress, reduce energy variance, and provide dependable forwarding under constrained duty cycles. The paper aims for neutral assessment of algorithmic behavior across diverse settings without overemphasizing any single improvement or metric.