Goal-Oriented Interference Coordination in 6G In-Factory Subnetworks

TL;DR

This work tackles the dense inter-subnetwork interference problem in 6G in-factory subnetworks by introducing CADIC, a decentralized, goal-oriented joint power and multi-sub-band allocation policy that uses the instantaneous control cost to steer radio resource use. The policy combines a logistic transmit power model and a rank-based, multi-sub-band selection mechanism, tuned via a multi-objective MOTPE Bayesian optimization to balance long-term plant stability and spectral efficiency. Compared with centralized BLER-optimized baselines, CADIC achieves over 2x higher subnetwork density while maintaining competitive control performance and reduced execution costs; a CADIC Modified variant further improves BLER in some regimes. The approach demonstrates the practical value of aligning wireless resource management with control objectives in dense industrial deployments, and points to future work on online learning and joint control-radio optimization.

Abstract

Subnetworks are expected to enhance wireless pervasiveness for critical applications such as wireless control of plants, however, they are interference-limited due to their extreme density. This paper proposes a goal-oriented joint power and multiple sub-bands allocation policy for interference coordination in 6G in-factory subnetworks. Current methods for interference coordination in subnetworks only focus on optimizing communication metrics, such as the block error rate, without considering the goal of the controlled plants. This oversight often leads to inefficient allocation of the limited radio resources. To address this, we devise a novel decentralized inter-subnetwork interference coordination policy optimized using a Bayesian framework to ensure the long-term stability of the subnetwork-controlled plants. Our results show that the proposed decentralized method can support more than twice the density of subnetwork-controlled plants compared to centralized schemes that aim to minimize the block error rate while reducing execution complexity significantly.

Figures (7)

• Figure 1: In-factory subnetworks supporting closed-loop control of their associated plants at the field level. The parent network is responsible for centralized radio resource allocation. However, the subnetworks may autonomously manage interference in a decentralized configuration.
• Figure 2: Illustration of the In-factory subnetwork control system model depicting the TDD frame structure for the periodic traffic, and the shared frequency-time resource grid with $L$ sub-bands, where each sub-band consists of $K$ channels.
• Figure 3: Analysis of the response of plant 1 and plant 2 to different inter-arrival times for perfect channel condition. Plant 2 requires a periodic closed-loop control at least every $4~\text{ms}$, while Plant 1 requires periodic closed-loop control at least every $2 ms$ to maintain a low control cost over a finite horizon.
• Figure 4: CADIC Policy for transmit power control and sub-band allocation. (a) Transmit power as a logistic function of instantaneous control cost, and parameters $k_0=0.49$ and $k_1=16$. (b) Number of sub-bands to be selected as a piecewise step function of the instantaneous control cost with parameters $z_1=100$ and $z_2=186$.
• Figure 5: Explanation of CADIC operation for a randomly selected plant wirelessly controlled by subnetwork. CADIC reduces the transmit power and number of operating sub-bands when the plant is more stable (low instantaneous control cost), and increases the transmit power and/or the number of sub-bands when the plant is less stable.

Theorems & Definitions (2)

• Remark 1
• Remark 2