Communications-Aware NMPC for Multi-Rotor Aerial Relay Networks Under Jamming Interference

In the sky above our cities and across remote landscapes, small flying machines are increasingly tasked with carrying out complex jobs, from inspecting power lines to extending cellular networks. These machines, known as multi-rotor aerial vehicles, rely on radio signals to talk to each other and to ground stations. However, their ability to stay connected is fragile. If a signal is blocked by a building or drowned out by a deliberate jammer—a device broadcasting noise to disrupt communication—the machine can lose its way or fail its mission. For decades, engineers have tried to solve this by making the radio signals themselves stronger or by having the machines move away from the noise. But these solutions often treat the machine's movement and its radio as separate problems. They assume the machine is just a point moving through space, ignoring a critical physical reality: the radio antenna is attached to the machine's body. When the machine tilts to move sideways, the antenna tilts with it. If the antenna is directional, meaning it sends and receives signals best in a specific direction, a simple tilt can point the antenna's "blind spot" directly at the person it needs to talk to, severing the link even if the machine is physically close.

This disconnect between how a machine moves and how its antenna points is the central challenge addressed by a new study from researchers at several European institutions. They developed a control system that treats the machine's flight path and its radio orientation as a single, unified problem. Instead of planning a route and then hoping the radio stays aligned, their system calculates the flight path specifically to keep the antenna pointed correctly while simultaneously dodging interference. The researchers tested this approach in computer simulations involving a relay scenario: one drone flying between a ground station and a second, mobile drone, while a jammer on the ground tried to cut off their connection. The results showed that by accounting for the antenna's directionality, the system could keep the connection alive and stable, whereas traditional methods that ignored the antenna's orientation caused the signal to drop to near-zero levels during maneuvers.

The core of the new approach is a two-part brain for the drone. The first part is a high-level planner that looks ahead and decides where the drone should go to keep the weakest link in the chain as strong as possible. It does not just look at distance; it considers the angle of the antenna relative to the other devices and the jammer. To make these complex calculations fast enough for real-time use, the researchers simplified the math by treating the antenna's signal strength as a geometric shape rather than a complex radiation pattern. This allows the planner to quickly figure out that a certain tilt might be bad for the signal and adjust the flight path accordingly. The second part is a low-level controller that actually moves the drone's motors. It takes the high-level plan and executes it, but with a crucial addition: it constantly checks if the drone's current tilt is ruining the signal. If the drone starts to lean too far, this controller gently corrects the flight to bring the antenna back into alignment, all while respecting the physical limits of the motors.

A key finding of the study is that the physical design of the drone matters just as much as the software. The researchers compared two types of drones: a standard one where all the propellers lie flat on a single plane, and a more advanced one where the propellers are tilted at an angle. In the standard design, moving sideways requires the entire drone to tilt, which inevitably points the antenna in the wrong direction. In the simulations, this design struggled to maintain a connection under jamming, often forcing the drone to fly much higher to find a geometry where the signal could survive. The tilted-propeller design, however, could move sideways without tilting its body, keeping the antenna pointed correctly. This mechanical advantage allowed the advanced drone to maintain a reliable connection with much less effort, proving that the hardware's ability to move independently of its orientation is a vital factor in surviving interference.

The simulations revealed that this integrated approach dramatically improved reliability. When the researchers compared their new method against older techniques that ignored antenna direction, the new system increased the minimum strength of the connection by nearly two orders of magnitude. In practical terms, this means the difference between a connection that drops out completely and one that stays strong enough to send data. While the average speed of data transfer improved only slightly, the most important gain was the elimination of sudden, catastrophic failures. The system proved robust even when the jammer turned on and off rapidly or when the drone's sensors had slight errors in locating the jammer. The controller did not overreact to these changes; instead, it maintained a steady, optimal position that worked for the worst-case scenario, ensuring the link never broke.

The study concludes that for aerial robots to operate reliably in hostile or crowded environments, engineers can no longer treat the radio and the flight mechanics as separate systems. The way a drone moves directly dictates the quality of its communication. By building control systems that understand this physical coupling, and by choosing drone designs that allow for independent movement and orientation, it is possible to keep these machines connected even when someone is actively trying to cut them off. The research suggests that while standard drones can be made to work, fully actuated designs with tilted propellers offer a significant advantage in maintaining the precise alignment needed for robust communication. This insight shifts the focus from simply making radios stronger to designing the entire vehicle as a communication-aware machine, where every movement is calculated to keep the signal alive.