Optimizing Collaborative UAV Networks for Data Efficiency in IoT Ecosystems

TL;DR

Problem $\mathsf{P}$, a strong NP-hard data-gathering task for collaborative UAV networks, is modeled to minimize energy while delivering multisource information to multiple destinations over time-varying connectivity. The authors formulate an ILP and develop scalable heuristics that route data via multiple UAV hubs using augmented virtual graphs, coupled with adaptive power control and dynamic resource allocation. They show that exact solutions are intractable for large instances and demonstrate that the MPF heuristic achieves near-optimal performance with ~99% speedup over ILP on smaller cases, while maintaining a deviation from the optimum around $14\%$. Experimental results indicate that energy consumption increases with packet size and decreases with bandwidth, and that larger UAV swarms improve load balancing and overall data-gathering efficiency in IoT deployments.

Abstract

Advances in the Internet of Things are revolutionizing data acquisition, enhancing artificial intelligence and quality of service. Unmanned Aerial Vehicles (UAVs) provide an efficient data-gathering solution across varied environments. This paper addresses challenges in integrating UAVs for large scale data operations, including mobility, multi-hop paths, and optimized multi-source information transfer. We propose a collaborative UAV framework that enables efficient data sharing with minimal communication overhead, featuring adaptive power control and dynamic resource allocation. Formulated as an NP-hard Integer Linear Program, our approach uses heuristic algorithms to optimize routing through UAV hubs. Simulations show promise in terms of computation time (99% speedup) and outcome (down to 14% deviation from the optimal).

Figures (8)

• Figure 1: System Model. The grey circles represent the sources of the data referred to.
• Figure 2: UAVs connectivity graph.
• Figure 3: The augmented graph of UAVs. $s_1$ is the virtual source of information $i$ and $d_{14}$, $d_{1|U|}$ are the virtual destinations of information $i$ that corresponds to UAV 4 and $|U|$, respectively.
• Figure 4: Algorithm performance
• Figure 5: The energy consumption level vs the packet size.

Theorems & Definitions (2)

• Proposition 1
• proof