The authors have developed an innovative topology by amalgamating consortium blockchain, often referred to as supervisory blockchain, with fog computing. The proposed system is organized into three distinct layers: the application layer, the fog layer, and the blockchain security layer. To accommodate this model effectively, the authors introduce the novel proof of enhanced concept (PoEC) consensus mechanism. This approach employs homomorphic encryption to secure transactions, which are then outsourced to the fog layer or fog devices. This strategy mitigates various security threats, including collusion attacks, phishing attacks, and replay attacks, bolstering the resilience of each layer against such incursions. To bolster security measures further, our model adopts a hybrid-deep learning protocol for safeguarding electronic medical records against potential breaches while concurrently reducing latency through a decentralized fog computing system.
TopIntroduction
In the realm of IoT (internet of things) and IoMT (internet of medical things), where data traffic is increasingly centralized within cloud-based systems, significant concerns have emerged (Al Hwaitat et al., 2023). These concerns encompass patient safety and confidentiality, as issues like data ownership, data collection, and location privacy are put at risk. One alarming vulnerability is the potential for hackers to target 5G-enabled IoMT networks by copying data and altering the identification of healthcare equipment. Currently, IoMT-Cloud systems face challenges related to single points of failure, malicious attacks, and privacy breaches (Mushtaq et al., 2016). To ensure the security of networks and the safe transmission of Personal Health Data (PHD), it is imperative that data exchanges between IoMT and the Cloud incorporate robust mechanisms for trust, device identification, and user authentication (UA) (Azaria et al., 2016). The safeguarding of individuals' private data has long been a focal point within the field of computer science. However, the rise of deep learning (DL) algorithms has, in some instances, led to their application without adequate consideration for data privacy. While these algorithms have delivered improved learning accuracy, privacy concerns have often been deferred. It was only with the deployment of remote DL models over the internet that a heightened interest in data privacy emerged. In response, various cloud platforms were developed, utilizing DL models for both training and inference phases. Subsequently, this service became known as Deep Learning as a Service (DLaaS) (Hasnain et al., 2020). Hence, the transmission and storage of private data over the internet pose significant risks, as servers holding this data are susceptible to vulnerabilities that could result in leakage. Such leaks may occur intentionally or unintentionally (Ali, Rahim, Pasha et al, 2021). Intentional breaches can happen through hacking, piracy, or social engineering tactics, whereas unintentional leaks may be attributed to users themselves or the employees of service providers. Remarkably, a substantial portion of data leakage incidents, totaling 43%, is attributed to employee actions, as reported by Intel Security (Ali & Mehboob, ; Ali et al., 2017; Hameed et al., 2016). Figure 1 illustrates the application of Fog and Cloud Computing, which aids planners in the establishment and delivery of services, thereby enhancing resource allocation and reducing service latency.
Figure 1. Application of fog and cloud environment
Top3. Research Contribution
This paper addresses the aforementioned challenges by introducing a novel blockchain-based deep-learning framework aimed at enhancing security and privacy within the context of supervisory blockchain-based fog computing. The primary contributions of this study encompass: