A Survey on Trustworthy Edge Intelligence: From Security and Reliability To Transparency and Sustainability

TL;DR

This survey defines trustworthy edge intelligence (EI) as a framework combining edge computing and AI to support real-time, secure, and transparent decision-making at the network edge. It presents a three-layer architecture (network, technology, application) and identifies four core characteristics—endogenous security, reliability, transparency, and sustainability—that jointly build trust in resource-constrained EI environments. It then surveys enabling technologies (security/privacy-preserving methods and interpretability) and tangible solutions (security, reliability, transparency, and sustainability) while detailing challenges such as heterogeneity, dynamic edge conditions, and the deployment of large models at the edge. The paper also outlines open issues and research directions, including zero-trust at the edge, generative-model-based security, autonomous edge co-inference, and data governance, aiming to guide future work toward robust, scalable, and trustworthy EI deployments with practical impact on latency-sensitive and privacy-sensitive edge applications.

Abstract

Edge Intelligence (EI) integrates Edge Computing (EC) and Artificial Intelligence (AI) to push the capabilities of AI to the network edge for real-time, efficient and secure intelligent decision-making and computation. However, EI faces various challenges due to resource constraints, heterogeneous network environments, and diverse service requirements of different applications, which together affect the trustworthiness of EI in the eyes of stakeholders. This survey comprehensively summarizes the characteristics, architecture, technologies, and solutions of trustworthy EI. Specifically, we first emphasize the need for trustworthy EI in the context of the trend toward large models. We then provide an initial definition of trustworthy EI, explore its key characteristics and give a multi-layered architecture for trustworthy EI. Then, we summarize several important issues that hinder the achievement of trustworthy EI. Subsequently, we present enabling technologies for trustworthy EI systems and provide an in-depth literature review of the state-of-the-art solutions for realizing the trustworthiness of EI. Finally, we discuss the corresponding research challenges and open issues.

Figures (6)

• Figure 1: Structure of the survey.
• Figure 2: Architecture of trustworthy EI: The network layer integrates communication, computation, perception, and intelligence, delivering the application layer high-performance, interpretable decision-making, resilient and sustainable solutions. Meanwhile, the technology layer establishes a trustworthy environment, offering robust technical support across the network and application layers, fostering seamless collaboration for comprehensive trustworthy EI.
• Figure 3: The multi-lateral trust model.
• Figure 4: Illustrative solutions of balanced security and privacy protection: a) Security solutions for wireless communications ensure availability of the edge network and security of data transmission; b) Security solutions for learning models address threats such as model poisoning and adversary attacks, ensuring trustworthiness and robustness of learning models; and c) Privacy-preserving solutions focus on handling sensitive data at the network edge.
• Figure 5: Illustrative solutions of reliability: a) Distributed collaboration: It allows heterogeneous devices to work together and share knowledge, reducing the risk of single points of failure while increasing resource utilization; and b) Incentive mechanisms: Rewards are given for honest and high-quality resource contributions.