Task-Specific Trust Evaluation for Multi-Hop Collaborator Selection via GNN-Aided Distributed Agentic AI

TL;DR

Problem: trustworthy multi-hop collaborator selection in dynamic, privacy-constrained networks. Approach: a GNN-aided distributed agentic AI (GADAI) that separately evaluates historical reliability and task-specific resource trust, then integrates results to plan value-maximizing multi-hop paths in a privacy-preserving, distributed manner. Contributions: a GNN-based historical reliability model, an LAM-enabled agentic AI system for resource trust and distributed path planning, and a comprehensive complexity and experimental validation showing improved VoC compared to baselines. Significance: enables scalable, privacy-aware collaboration across IoT, vehicular, and edge networks, improving task completion success in dynamic topologies.

Abstract

The success of collaborative task completion among networked devices hinges on the effective selection of trustworthy collaborators. However, accurate task-specific trust evaluation of multi-hop collaborators can be extremely complex. The reason is that their trust evaluation is determined by a combination of diverse trust-related perspectives with different characteristics, including historical collaboration reliability, volatile and sensitive conditions of available resources for collaboration, as well as continuously evolving network topologies. To address this challenge, this paper presents a graph neural network (GNN)-aided distributed agentic AI (GADAI) framework, in which different aspects of devices' task-specific trustworthiness are separately evaluated and jointly integrated to facilitate multi-hop collaborator selection. GADAI first utilizes a GNN-assisted model to infer device trust from historical collaboration data. Specifically, it employs GNN to propagate and aggregate trust information among multi-hop neighbours, resulting in more accurate device reliability evaluation. Considering the dynamic and privacy-sensitive nature of device resources, a privacy-preserving resource evaluation mechanism is implemented using agentic AI. Each device hosts a large AI model-driven agent capable of autonomously determining whether its local resources meet the requirements of a given task, ensuring both task-specific and privacy-preserving trust evaluation. By combining the outcomes of these assessments, only the trusted devices can coordinate a task-oriented multi-hop cooperation path through their agents in a distributed manner. Experimental results show that our proposed GADAI outperforms the comparison algorithms in planning multi-hop paths that maximize the value of task completion.

Figures (11)

• Figure 1: A task is transmitted from the task owner through a sequence of trusted terminal devices to a trusted edge computing device.
• Figure 2: The proposed GADAI framework. (1) Historical reliability evaluation of devices involves four steps: historical collaboration graph construction, trust propagation and aggregation, historical reliability computation, and the removal of devices below the trust thresholds. (2) Task-specific device resource trust assessment and multi-hop path planning enabled by the agentic AI system.
• Figure 3: Task-specific resource trust evaluation for an EC device using OpenAI-o3-mini.
• Figure 4: Training loss comparison for different learning rates.
• Figure 5: Comparison of results with varying minimum trust demand $c^{\text{TF}}$. (a) Comparison of the average VoC values. As $c^{\text{TF}}$ increases, the average VoC values decrease for all methods, while GADAI consistently achieves higher values than the comparison methods. (b) Comparison of the number of trusted terminal devices. As $c^{\text{TF}}$ increases, the number of trusted terminal devices for all methods decreases.

Theorems & Definitions (3)

• Definition 1: Task forwarding trust
• Definition 2: Task computing trust
• Definition 3: Task-specific trusted multi-hop collaboration path