Unlocking Sub-THz by Robotic Aerial Base Stations: Joint Deployment and Wireless Backhaul Routing

Despite the numerous advantages of aerial base stations (ABSs), including their relatively ease of deployment and inherent flexibility for relocation to adapt to highly dynamic networks, their service endurance is constrained by the limited capacity of their onboard batteries. To address this limiting factor, we explore the use of robotic aerial base stations (RABSs) that are equipped with grasping end-effectors able to anchor onto tall urban landforms such as lampposts. Energy-neutral anchoring conserves energy consumption by eliminating the need for hovering or flying during service time, thereby massively improving communication service endurance. In this paper, a joint RABS deployment and wireless backhauling scheme with the aim of maximizing served traffic is proposed to support future dynamic and densified wireless networks experiencing unprecedented data traffic growth. To meet this significant increase in traffic demand, which requires substantial bandwidth for both access and backhaul, we employ sub-Terahertz (sub-THz) band communication due to its ultra-wide spectrum resources. Given the sub-THz band’s susceptibility to blockages and severe propagation losses due to absorption, we propose a multi-hop wireless scheme to extend network coverage. The optimization interplay between RABS grasping locations, route flow control, and sub-band allocation to ensure link capacity, is framed as a robust optimization problem aimed at maximizing served traffic with a cardinality-constrained uncertainty set. Since the grasping locations are determined from all candidate locations, the number of corresponding candidate routes can significantly increase with the network size in this multi-hop enabled network. In this work, we propose a column generation (CG) based algorithm to tackle the curse of dimensionality due to the exponentially increased number of candidate routes. To this end, a near-optimal decision making is proposed with significantly reduced computational complexity. A wide set of numerical investigations demonstrates the superiority of the proposed network scheme over baseline schemes. For instance, the aggregated served traffic demand improved by 125% compared to a network with fixed small cell deployment which could be considered as the nominal use case and a common deployment option for increasing network capacity.