Edge Information Hub: Orchestrating Satellites, UAVs, MEC, Sensing and Communications for 6G Closed-Loop Controls

Chengleyang Lei, Wei Feng, Wei Peng, Yunfei Chen, Ning Ge, Shiwen Mao

TL;DR

This paper introduces the Edge Information Hub (EIH), a UAV-mounted system designed to manage field robots in remote or disaster-stricken areas. By jointly optimizing sensing, computing, and communication resources—and balancing processing between local edge servers and satellite-linked clouds—the authors significantly reduce the "control cost" (LQR) of multi-robot systems compared to traditional communication-focused methods.

What Problem Does This Paper Address?

In disaster recovery or remote exploration, field robots often lack the onboard processing power or sensor range to operate autonomously. They rely on external "helpers" for sensing the environment, computing control commands, and communicating those instructions.

Traditionally, these three pillars—Sensing, Communication, and Computing—are designed in silos. However, in the era of 6G, these functions are deeply coupled in what the authors call $SC^3$ loops (Sensing-Communication-Computing-Control). The challenge is that the "helper" (often a UAV) has limited battery and processing power, while the alternative (offloading to a remote cloud via satellite) introduces significant latency. This paper tackles the problem of how to orchestrate these limited resources to ensure robots perform their tasks accurately and efficiently without the system becoming unstable.

Methods

The authors model the system as a set of multiple closed-loop controls. To measure performance, they use the Linear Quadratic Regulator (LQR) cost, a standard metric in control theory that penalizes both the robot's deviation from its target state and the energy spent on control.

The core of their approach is a joint optimization problem that manages:

  1. Transmit Power: How much energy the UAV uses to send commands to robots.
  2. Computing Allocation: How the UAV’s Mobile Edge Computing (MEC) server divides its CPU cycles.
  3. Satellite Backhaul: How much data is sent to the remote cloud.
  4. Data Splitting: A novel three-way split where sensor data is either processed entirely on the UAV, pre-processed on the UAV then sent to the cloud, or sent entirely to the cloud.

Because the resulting optimization problem is "non-convex" (notoriously difficult to solve directly), the authors first derive an analytical solution for the optimal data splitting ratio. They then apply Successive Convex Approximation (SCA), an iterative algorithm that simplifies the complex problem into a series of manageable ones, converging on a sub-optimal but highly effective solution.

Datasets

The paper does not use a pre-existing real-world dataset. Instead, the authors evaluate their approach using numerical simulations in MATLAB.

  • Environment: A circular area with a 5000m radius containing 5 randomly distributed robots.
  • Parameters: Standard 6G-relevant values, such as 50x50 state matrices, LEO satellite heights (1500 km), and specific noise variances for Gaussian sensing and control noise.

Key Insights and Ideas

  • The End of Silos: The biggest "aha" moment is that a communication-oriented scheme (which simply tries to maximize data throughput) actually performs the worst for control tasks. In some cases, it can even lead to system instability where the proposed closed-loop-oriented scheme remains stable.
  • The Three-Way Split: Unlike many edge computing papers that offer a binary "local vs. cloud" choice, this paper shows that "pre-processing" at the edge before sending to the cloud is often the "sweet spot" for minimizing latency in $SC^3$ loops.
  • Information Entropy as a Bridge: The authors use information entropy to link communication quality directly to control performance. This allows them to define a "minimum entropy" required to achieve a specific control accuracy.
  • Diminishing Returns: There is a clear saturation point. Once transmit power or computing capability reaches a certain threshold, the LQR cost hits a floor determined by the intrinsic noise of the sensors.

Key Results

  • Rapid Convergence: The proposed iterative algorithm is highly efficient, typically converging within just three iterations, making it viable for real-time deployment.
  • Superior Control: In simulations, the proposed scheme consistently achieved a lower LQR cost than benchmarks. For instance, at lower transmit power levels (e.g., 9 dBW), the proposed scheme maintained stability while traditional communication-oriented schemes saw costs spike toward infinity.
  • Resource Sensitivity: The results quantify how sensing noise variance degrades control. Interestingly, the authors show that increasing communication power can partially compensate for poor sensor quality, but only up to the limits imposed by the system's "intrinsic entropy."

Limitations and Future Work

  • Interference Neglect: The current model assumes orthogonal channels, meaning robots don't interfere with each other's signals. In dense robot swarms, managing inter-robot interference will be a critical next step.
  • Gaussian Assumptions: The mathematical proofs rely on Gaussian noise distributions. While the authors suggest extensions for non-Gaussian noise, a more robust framework for "messy" real-world noise is needed.
  • Multi-Robot Cooperation: The paper treats each robot's control loop as independent. Future work could explore "cooperative $SC^3$," where robots share data to complete a single, complex task.

Summary updated

Abstract

An increasing number of field robots would be used for mission-critical tasks in remote or post-disaster areas. Due to the limited individual abilities, these robots usually require an edge information hub (EIH), with not only communication but also sensing and computing functions. Such EIH could be deployed on a flexibly-dispatched unmanned aerial vehicle (UAV). Different from traditional aerial base stations or mobile edge computing (MEC), the EIH would direct the operations of robots via sensing-communication-computing-control ($\textbf{SC}^3$) closed-loop orchestration. This paper aims to optimize the closed-loop control performance of multiple $\textbf{SC}^3$ loops, with constraints on satellite-backhaul rate, computing capability, and on-board energy. Specifically, the linear quadratic regulator (LQR) control cost is used to measure the closed-loop utility, and a sum LQR cost minimization problem is formulated to jointly optimize the splitting of sensor data and allocation of communication and computing resources. We first derive the optimal splitting ratio of sensor data, and then recast the problem to a more tractable form. An iterative algorithm is finally proposed to provide a sub-optimal solution. Simulation results demonstrate the superiority of the proposed algorithm. We also uncover the influence of $\textbf{SC}^3$ parameters on closed-loop controls, highlighting more systematic understanding.
BibTeX
@misc{lei2024edge,
  author = {Chengleyang Lei and Wei Feng and Wei Peng and Yunfei Chen and Ning Ge and Shiwen Mao},
  title = {{Edge Information Hub: Orchestrating Satellites, UAVs, MEC, Sensing and Communications for 6G Closed-Loop Controls}},
  year = {2024},
  eprint = {2403.06579},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2403.06579},
}
Figure