Traditional cloud-centric approaches to medical data sharing pose risks related to real-time performance, security, and stability. Medical and healthcare data encounter challenges like data silos, privacy breaches, and transmission latency. In response to these challenges, this paper introduces a blockchain-based framework for trustworthy medical data sharing in edge computing environments. Leveraging healthcare consortium edge blockchains, this framework enables fine-grained access control to medical data. Specifically, it addresses the real-time, multi-attribute authorization challenge in CP-ABE through a Distributed Attribute Authorization strategy (DAA) based on blockchain. Furthermore, it tackles the key security issues in CP-ABE through a Distributed Key Generation protocol (DKG) based on blockchain. To address computational resource constraints in CP-ABE, we enhance a Distributed Modular Exponentiation Outsourcing algorithm (DME) and elevate its verifiable probability to “1”. Theoretical analysis establishes the IND-CPA security of this framework in the Random Oracle Model. Experimental results demonstrate the effectiveness of our solution for resource-constrained end-user devices in edge computing environments.

Blockchain, as a distributed trust database, has been widely applied in the field of trustworthy sharing of Internet of Things (IoT) sensor data. A single hash mechanism has achieved, to some extent, the trustworthy on-chain storage of blockchain sensor data, that is, the consistency of data on and off the chain. However, it still faces potential security risks such as collision attacks, short password attacks, and rainbow table attacks. To address this issue, this paper proposes a resiliently secure blockchain sensor data trustworthy sharing model based on a mimic hash mechanism. Specifically, in response to the security risks that may arise from the single hash mechanism, this study innovatively introduces a mimic hash mechanism and proposes two methods for constructing mimic hashes based on Verifiable Random Function (VRF) and Cyber Mimic Defense (CMD) in dedicated Wireless Sensor Networks (WSNs) and open public networks, respectively. Theoretical analysis and experimental results demonstrate that this model effectively solves the problem of trustworthy on-chain storage of sensor data in edge computing environments, enhancing the trustworthiness and security of the data on the chain.

Indiscriminate data poisoning attacks are highly effective against unsupervised learning. However, recent studies show that contrastive learning is also susceptible to data poisoning attacks. As a form of data poisoning attack, the attacker adds poison to the clean pre-training dataset. This article proposes IDPA, an indiscriminate data poisoning attack targeting the encoder in contrastive learning. where the attacker’s goal is to directly poison the pre-trained encoder. The feature vectors of any clean sample and the attacked sample from the attacker will exhibit high similarity, causing the downstream classifier to misclassify the clean sample as the samples designated by the attacker. Therefore, this article formulates IDPA as a dual optimization problem and defines two loss functions: the attack effectiveness loss and the model utility loss. These losses are associated with effectively poisoning the pre-trained encoder and maintaining the accuracy of the downstream classifier, respectively. During training, the attack affects the contrastive learning algorithm and predictions are made on multiple datasets. Experimental results show that the attack success rate of 92%. This article evaluates the effectiveness of IDPA on the CLIP dataset released by OpenAI, with attack success rate of 88%.