Metaheuristic Optimization of Trajectory and Dynamic Time Splitting for UAV Communication Systems

TL;DR

This work tackles non-convex sum-rate optimization for a UAV-enabled wireless system where the UB employs backscatter communication and local caching. It jointly optimizes the UAV trajectory $\pmb{\gamma}$ and dynamic time-splitting $\delta_i$ over $N$ slots to maximize the end-user rate while satisfying energy harvesting, mobility, and data-rate constraints, under imperfect CSI with a Doppler-aware channel model. To solve the challenging problem, the authors propose two metaheuristics: a Genetic Algorithm (GA) and an Improved Particle Swarm Optimization (IPSO), each with tailored encoding, mutation/crossover/selection strategies, and computational complexities $O(S \log(S) G (3N+3)^2)$ and $O(S G (3N+3)^2)$, respectively. Numerical results show that IPSO excels at low WPT power and GA performs better at higher power, with both approaches achieving higher rates and lower computation times than BCD and standard PSO baselines. The findings demonstrate the practical viability of integrating UAVs with backscatter and caching to enhance throughput in energy-constrained wireless networks.

Abstract

The integration of unmanned aerial vehicles (UAVs) into wireless communication systems has emerged as a transformative approach, promising cost-efficient connectivity. This paper addresses the optimization of the dynamic time-splitting ratio and flight trajectory for a communication system linking a ground base station to the UAV equipped with backscatter devices (referred to as UB), and from UB to an end user. Given the inherent non-convexity of the problem, we develop two meta-heuristic-based approaches inspired by genetic algorithm and particle swarm optimization to enhance the total achievable rate while reducing computational complexity. Numerical results demonstrate the effectiveness of these meta-heuristic solutions, showcasing significant improvements in the achievable rate and computation time compared to existing benchmarks.

Figures (5)

• Figure 1: Illustration of the system model, where the GBS serves the end user via a UB.
• Figure 2: Achievable rate versus transmit power of GBS and UB.
• Figure 3: Achievable rate v.s. flying time and altitude of UB.
• Figure 4: Analysis of computation time and complexity.
• Figure 5: Trajectory of UAV and dynamic time splitting.