Abstract
The proliferation of novel industrial applications at the wireless edge, such as smart grids and vehicle networks, demands the advancement of cyber-physical systems (CPS). The performance of CPSs is closely linked to the last-mile wireless communication networks, which often become bottlenecks due to their inherent limited resources. Current CPS operations often treat wireless communication networks as unpredictable and uncontrollable variables, ignoring the potential adaptability of wireless networks, which results in inefficient and overly conservative CPS operations. Meanwhile, current wireless communications often focus more on throughput and other transmission-related metrics instead of CPS goals. In this study, we introduce the framework of goal-oriented wireless communication resource allocations, accounting for the semantics and significance of data for CPS operation goals. This guarantees optimal CPS performance from a cybernetic standpoint. We formulate a bandwidth allocation problem aimed at maximizing the information utility gain of transmitted data brought to CPS operation goals. Since the goal-oriented bandwidth allocation problem is a large-scale combinational problem, we propose a divide-and-conquer and greedy solution algorithm. The information utility gain is first approximately decomposed into marginal utility information gains and computed in a parallel manner. Subsequently, the bandwidth allocation problem is reformulated as a knapsack problem, which can be further solved greedily with a guaranteed sub-optimality gap. We further demonstrate how our proposed goal-oriented bandwidth allocation algorithm can be applied in four potential CPS applications, including data-driven decision-making, edge learning, federated learning, and distributed optimization.
Introduction: Beyond the Shannon Paradigm
In modern industrial applications, the "last-mile" wireless link is frequently the primary performance bottleneck. Traditionally, wireless engineering has been governed by Shannon’s paradigm, which prioritizes maximizing bits per second (throughput) or minimizing bit error rates. However, for a Cyber-Physical System (CPS) operator, not all bits are created equal. For instance, a sensor reading during a power grid failure is significantly more valuable than a routine status update, even if both data packets are identical in size.
Current CPS operations often treat wireless networks as unpredictable "black boxes," leading to inefficient designs. Conversely, standard communication protocols ignore the semantics (meaning) and significance of the data they carry. This research addresses this gap by creating a mathematical bridge between communication resources (bandwidth) and physical system performance, ensuring that communication serves the ultimate system goal.
Methodology: Maximizing Information Utility
The core of this framework is the maximization of Information Utility Gain. Because allocating bandwidth across thousands of devices is a large-scale combinatorial problem (NP-hard), the authors introduce a three-step solution:
- Divide-and-Conquer: The framework leverages the submodular property (the law of diminishing returns) inherent in most CPS goal functions. This allows the decomposition of a complex global objective into "marginal utility gains" for each individual device.
- Knapsack Reformulation: The bandwidth allocation is transformed into a variation of the "Knapsack Problem," where the "weight" represents the required radio resource blocks and the "value" represents the marginal utility of a device's data.
- Greedy Algorithm: A fast, greedy solution with a complexity of $O(J \log J)$ is proposed for real-time execution on edge servers. This includes a "Hybrid" rule that balances data importance with wireless channel quality.
To address the "Causality Problem"—the challenge of knowing data utility before it is received—the system uses two strategies: for large payloads (e.g., images), devices upload a small metadata "utility value" first; for small payloads, the server predicts utility based on historical trends.
A Unified Framework for CPS Applications
The strength of this approach lies in its versatility. The paper provides a mathematical "recipe book" for calculating marginal utility across four distinct industrial tasks: data-driven decision-making, edge learning, federated learning, and distributed optimization.
Key Results and Performance Gains
The proposed "Hybrid" goal-oriented approach consistently outperformed traditional channel-based methods across various simulations, including MNIST classification, smart grid load shedding, and vehicle routing:
- Decision Making: Reduced decision costs by approximately 50% compared to traditional policies under the same communication budget.
- Edge Learning: Reached 95% accuracy in roughly half the time (50% fewer data collection phases) compared to standard methods.
- Federated Learning: Achieved target accuracy with a 30% reduction in total training time.
- Distributed Optimization: Reached the optimality gap ($10^{-3}$) 30% faster by prioritizing devices with the most significant local variable changes.
Limitations and Future Outlook
While the framework shows significant promise, current models omit small-scale fading effects to simplify allocation logic. Future research will focus on integrating realistic, fast-fading channel models and exploring the "co-design" of physical control laws and communication protocols simultaneously. Additionally, integrating "hard" real-time latency constraints into the utility framework remains a critical next step for safety-critical CPS tasks.
Summary updated