This article focuses on zero-trust foundation models (ZTFMs), a novel paradigm that embeds zero-trust security principles into the lifecycle of foundation models (FMs) for Internet of Things (IoT) systems. By integrating core tenets, such as least privilege access, continuous verification, data confidentiality, and behavioral analytics into the design, training, and deployment of FMs, ZTFMs can enable secure, privacy-preserving artificial intelligence (AI) across distributed, heterogeneous, and potentially adversarial IoT environments. We present the first structured synthesis of ZTFMs, identifying their potential to transform conventional trust-based IoT architectures into resilient, self-defending ecosystems. Moreover, we propose a comprehensive technical framework, incorporating federated learning (FL), blockchain-based identity management, micro-segmentation, and trusted execution environments (TEEs) to support decentralized, verifiable intelligence at the network edge. In addition, we investigate emerging security threats unique to ZTFM-enabled systems and evaluate countermeasures, such as anomaly detection, adversarial training, and secure aggregation. Through this analysis, we highlight key open research challenges in terms of scalability, secure orchestration, interpretable threat attribution, and dynamic trust calibration. This survey lays a foundational roadmap for secure, intelligent, and trustworthy IoT infrastructures powered by FMs.

Cooperative localization can improve both the availability and accuracy of positioning systems, and distributed belief propagation is a promising enabling technology. Difficulties with belief propagation lie in achieving high accuracy without causing high communication overhead and computational complexity. This limits its application in practical systems with mobile nodes that have limited battery size and processing capabilities. In this paper, we propose an efficient cooperative localization algorithm that can be applied to a real indoor localization system with a nonGaussian ranging error distribution. We first propose an asymmetric double exponential ranging error model based on empirical ranging data. An efficient cooperative localization algorithm based on distributed belief propagation is then proposed. The communication and computational cost is reduced by passing approximate beliefs represented by Gaussian distributions between neighbours and by using an analytical approximation to compute peer-to-peer messages. An extension of the proposed algorithm is also proposed for tracking dynamic nodes. The proposed algorithms are validated on an indoor localization system deployed with 28 nodes covering 8000 m², and are shown to outperform existing algorithms. In particular, the fraction of nodes located to one-meter accuracy is doubled using the proposed ranging error model and localization algorithm.

We study the problem of locating semi-static targets in non-line-of-sight (NLOS) environments using multiple sets of time-difference-of-arrival (TDOA) measurements. Specific focus is placed on the correlation between TDOA measurements caused by NLOS errors. We propose an algorithm which takes this correlation into account when fusing the available TDOA measurements and achieves high accuracy in estimating both the position and the associated uncertainty. In addition, a statistical ranging error model for NLOS environments is proposed based on empirical ranging data, which substantially improves the positioning accuracy. The proposed approach is verified experimentally in an aircraft cabin and achieves seat-level accuracy for 99% of the nodes.

The performance of a range-based indoor positioning system is severely degraded by non-line-of-sight (NLOS) propagation due to the offsets in range measurements (i.e. NLOS errors). It is difficult to predict or mitigate the NLOS errors since they are dependent on both the location and the environment. In this paper, we propose an accurate tracking scheme for NLOS environments by jointly estimating the target's trajectory and the NLOS errors based on the fusion of sensors that measure the motion of the target. We first formulate a MAP estimation problem with generic equality constraints that capture the spatial correlation of NLOS errors. A specific constraint function based on Gaussian process (GP) regression is then provided, and an iterative algorithm is proposed to solve the optimization problem. The proposed scheme is validated experimentally in an indoor positioning system with 125 MHz bandwidth using a mobile node equipped with an inertial measurement unit (IMU). It is shown that the median positioning error in an office environment is reduced by 90% to 11 cm compared to using conventional tracking algorithms.