Logic-Driven Semantic Communication for Resilient Multi-Agent Systems

TL;DR

The paper tackles the challenge of resilience in large-scale decentralized MAS for 6G by proposing a principled two-dimensional framework: epistemic resilience (sensing, knowledge repair, and mutual alignment) and action resilience (recovery and maintenance of optimal policies). It formalizes these notions using temporal epistemic logic and Kripke structures, and provides recoverability and durability metrics to quantify resilience. A semantic-communication-enabled agent architecture and decentralized policies are developed to achieve these guarantees, with formal verification ensuring bounded-horizon soundness and completeness. Through a Kripke-based learning framework and a distributed bandit case study, the work demonstrates faster and more stable recovery under abrupt stressors, highlighting the value of knowledge-driven coordination for resilient decentralized MAS in next-generation networks.

Abstract

The advent of 6G networks is accelerating autonomy and intelligence in large-scale, decentralized multi-agent systems (MAS). While this evolution enables adaptive behavior, it also heightens vulnerability to stressors such as environmental changes and adversarial behavior. Existing literature on resilience in decentralized MAS largely focuses on isolated aspects, such as fault tolerance, without offering a principled unified definition of multi-agent resilience. This gap limits the ability to design systems that can continuously sense, adapt, and recover under dynamic conditions. This article proposes a formal definition of MAS resilience grounded in two complementary dimensions: epistemic resilience, wherein agents recover and sustain accurate knowledge of the environment, and action resilience, wherein agents leverage that knowledge to coordinate and sustain goals under disruptions. We formalize resilience via temporal epistemic logic and quantify it using recoverability time (how quickly desired properties are re-established after a disturbance) and durability time (how long accurate beliefs and goal-directed behavior are sustained after recovery). We design an agent architecture and develop decentralized algorithms to achieve both epistemic and action resilience. We provide formal verification guarantees, showing that our specifications are sound with respect to the metric bounds and admit finite-horizon verification, enabling design-time certification and lightweight runtime monitoring. Through a case study on distributed multi-agent decision-making under stressors, we show that our approach outperforms baseline methods. Our formal verification analysis and simulation results highlight that the proposed framework enables resilient, knowledge-driven decision-making and sustained operation, laying the groundwork for resilient decentralized MAS in next-generation communication systems.

Figures (9)

• Figure 1: Proposed agent architecture illustrating two agents communicating with each other to refine their internal models. Dashed communicative arrows indicate that communication is not necessarily continuous and may occur as needed. Agent $j$ has a similar structure for its 'Ìnternal Processing' unit as the one depicted in (b).
• Figure 2: Example environment and its corresponding Kripke structure. In (a), agent $1$ (green) is placed on cell $1$, and agent $2$ (orange) on cell $2$. In (b), nodes represent possible worlds, with green and orange edges denoting the accessibility relations of agents $1$ and $2$, respectively. Reflexive arrows are omitted for clarity.
• Figure 3: Example environment and the corresponding per-agent slices of the shared Kripke structure, shown before and after communication. An agent’s slice is the triple $(W,R_{i,t},v)$ specific to agent $i$. Nodes represent possible worlds, and edges depict the agent-specific accessibility relations. Reflexive self-loops at worlds are omitted for clarity.
• Figure 4: Example environment after an environmental change (stressor) and the corresponding per-agent slices of the shared Kripke structure, shown after each agent executes the revise epistemic action. An agent’s slice is the triple $(W,R_{i,t},v)$ specific to agent $i$. Nodes represent possible worlds, and edges depict the agent-specific accessibility relations. Reflexive self-loops at worlds are omitted for clarity.
• Figure 5: Flow diagram of the epistemic policy executed by every agent to achieve epistemic resilience.

Theorems & Definitions (17)

• Example 1: Kripke structure for a two-agent grid world
• Remark 1: Intuitive mapping between formal logic and operational algorithms
• Remark 2: Collisions as stressors
• Example 2: Epistemic state refinement through communication
• Example 3: Epistemic state revision under a stressor
• Definition 1: System Resilience
• Definition 2: Epistemic Resilience
• Definition 3: Epistemic Recoverability
• Definition 4: Epistemic Durability
• Definition 5: Action Resilience