Abstract
This paper introduces a novel framework that leverages Simultaneously Transmitting and Reflecting Reconfigurable Intelligent Surfaces (STAR-RIS) to enhance the Internet of Robotic Things (IoRT). By integrating sensing, computing, and communication (ISCC), the system enables a base station to track target robots while simultaneously processing computational tasks offloaded by decision robots. We formulate a computation rate maximization problem and solve it using an alternating optimization algorithm. Numerical results demonstrate that STAR-RIS provides superior 360-degree coverage and higher data-processing capacity compared to conventional RIS technologies.
Introduction
The Internet of Robotic Things (IoRT) represents a demanding frontier where robots must simultaneously sense their environment, communicate with networks, and perform complex computations for decision-making. Due to hardware constraints like limited battery life and processing power, robots often rely on Mobile Edge Computing (MEC) to offload tasks to a nearby base station (BS). However, two major hurdles persist: signal blockages in complex environments and the resource competition between sensing and communication tasks.
While standard Reconfigurable Intelligent Surfaces (RIS) can bypass obstacles by reflecting signals, they are limited to a 180-degree half-space. This paper proposes the use of STAR-RIS, which can both reflect and transmit signals, providing full 360-degree coverage and significantly more degrees of freedom for network optimization.
System Model and ISCC Framework
The STAR-RIS-aided ISCC system model illustrating the transmission and reflection regions for decision robots and the sensing of a target robot. Source, p. 3
The proposed framework involves a Full-Duplex (FD) base station and a STAR-RIS assisting a group of "Decision Robots" (DRs) in offloading tasks while the BS tracks a "Target Robot" (TR). The STAR-RIS partition allows the BS to serve robots located in both the reflection and transmission regions simultaneously, overcoming the spatial limitations of conventional RIS.
Proposed Optimization Methodology
Internal architecture of the Full-Duplex Base Station, showing the separate Tx and Rx chains for simultaneous sensing and computation offloading. Source, p. 4
To maximize the network's computation rate, the authors optimize sensing beamformers, receive beamformers, and STAR-RIS coefficients. The BS architecture utilizes separate antenna arrays for transmission (Tx) and reception (Rx) to facilitate full-duplex operation. The optimization problem is non-convex and is addressed using an Alternating Optimization (AO) algorithm:
- Beamforming Optimization: Employs the Weighted Minimum Mean-Square Error (WMMSE) method to shape sensing signals and filter incoming offloading data.
- STAR-RIS Optimization: Uses a Penalty Method combined with Successive Convex Approximation (SCA) to manage the physical constraints of the surface elements.
Performance Evaluation and Key Insights
The framework was evaluated through numerical simulations using Rician fading channel models. Key findings include:
- 360-Degree Advantage: STAR-RIS consistently outperforms conventional RIS by serving robots on both sides of the surface, leading to higher sum computation rates.
- Sensing-Computing Trade-off: There is an inherent conflict between sensing accuracy and computing throughput. As sensing requirements (SINR) increase, the BS must divert power from MEC servers, resulting in a lower total computation rate.
- Scalability: Increasing the number of STAR-RIS elements significantly boosts performance, widening the gap over older technologies.
Conclusion and Future Work
The STAR-RIS-aided ISCC framework offers a promising solution for high-performance IoRT networks. While the current study assumes perfect interference cancellation and stationary robots, future research will explore high-mobility scenarios and the integration of downlink communication to complete the ISCC loop. These advancements will be crucial for real-world deployment where hardware imperfections and dynamic environments are prevalent.
Summary updated