Figure 1.8 IoT architecture for healthcare.
The overall objective was to minimize the effects of varying latency and bandwidth between gateways and servers here; it is the traditional cloud computing. It is tested with hierarchical computing architecture where existing machine learning methods can help in a fog-enabled IoT system. At a local level, it explores the feasibility of delivering adaptive transmission of data inside a closed-loop environment [27].
1.5.3 Developing Autonomous Capability is Key for Remote Healthcare Management
Remote monitoring is a robust model of healthcare. The advent of wearable devices aided in the smart healthcare application, in turn, enabling quicker clinical diagnosis and ease of disease prediction. The host of sensors is deployed in healthcare, uniquely placed inside the body either as implants or sensing devices over the physical body. Many of these can also communicate with small handheld devices such as smartphones or digital assistants. Dynamic and Interoperable Communication Framework (DICF) is primarily designed [28] to improve efficiency and enhance the decision-making capabilities of these wearable sensors. It optimizes various constraints of a sensor, namely, lifetime, storage capacity, handling multiple communication channels, and decision-making. The radio transmitter is used by sensors to interact with Aggregator Mobile Device (AMD); it can be a mobile platform. Information received is distinguished by the physiological source or based on sensor types. Periodic sensing for the collection of data varies with the kind of sensor and the human body. This periodic sensing is called a session. A session is, in turn, capable of sensing both periodic and event while monitoring is in progress. Pattern and range are clearly defined for transmitting data from AMD to service provider or clinic. So, when their data which does not fall within this range, an abnormality detected. AMD quickly generates an emergency notification and prioritizes transmission. In such events of AMD transmits completed sensed data to the clinic or the service provider without requesting the permission of end-users or patients.
Before the trigger of emergency is actuated, local machine learning algorithm checks for early diagnosis and first aid suggestions. Machine learning starts to establish causes and conditions using sensory data with varied conditions of patients and different patient data. Data collected and analyses at a local level significantly increase as different learning algorithms are applied to the results for further analysis. AMD performs the role of the hub by collecting data and relaying it to the clinic of the service provider, data accumulation at periodic intervals, and data gathering post applying algorithms. With the capacity of AMD, there is a significant improvement in processing capability at the local level. Data is stored as an Electronic Medical Record (EMR), which is interconnected with the notification system and monitoring unit. This enables a process of classification and decisions based on regression models. This decision-making model improves the performances of wearable devices in terms of event detection and emergency interval identification in a remote monitoring system. DICF aids in building a robust sensor dependent personal healthcare system. It also facilitates data collection, event detection, analysis, and communication with interconnected wearable sensors. AI-based remote healthcare system is shown in Figure 1.9.
Figure 1.9 AI-based remote healthcare management.
1.5.4 Enabling Data Privacy and Security in the Field of Remote Healthcare Management
Internet of Medical Things (IoM) has gained specialization in recent years with growing applications. Both embedded and wearable sensors gather comprehensive information about patients. It is shared with medical professionals for diagnosis. The sensitive nature of this data gives rise to protection and privacy, which is currently a significant challenge of the IoM. It uses anonymity-based authentication to mitigate privacy issues. To ensure the session is secure, upon mutual authentication, both medical professionals and medical sensors utilize private session keys. Research in this area of authenticity and privacy of data is still at the nascent stage. Currently, available authentication features are not suitable to achieve the privacy goals in terms of its features. At this juncture, there is not a credible and efficient authentication program in this segment. One of the key issues in achieving the goal is two-factor security in the event of loss or tampered smart card. Research is deplored to investigate the adversarial model, which is expected to mitigate various redundancies and ambiguities. This paper explores a methodology with 12 independent criteria analyzed using an adversary model for practical use. Broadly, it enables a better understanding of privacy requirements if not successful. In [28], the authors explore the feasibility of smart revocation/reissue and improve security efficiencies using a formal model. Secure-Anonymous Biometric-Based User Authentication Scheme (SAB-UAS) is tested for efficiency and meeting security goals.
In [29], the authors present SAB-UAS using a smart card with three entities in the healthcare communication chain, medical practitioners, wireless gateways, and wearable sensors. Moment SAB-UAS scheme starts process, two master keys are generated along with a long-term secret key by Wireless gateways, which is then transmitted to wearable sensors. Gateways quickly try to use one of the master keys, establishing it as a public key. This system is simulated at three stages, mainly user name registration, system login and authentication, and any case of revocations. After simulating SAB-UAS at three stages, the formal security analysis is also simulated. Random Oracle model is used to prove the security efficiency of SAB-UAS. The simulation clearly shows SAB-UAS scheme can securely protect sensitive information from various retrieval mechanisms. An informal security attack is also simulated with different 12 independent conditions, and it can meet the security goals setup. It is essential to understand whether SAB-UAS is also efficient because the electronic healthcare system comprises lightweight resources with very many limitations like storage capacity, bandwidth, and processing capabilities. Resource efficiency analysis is computed to understand storage, communication, and computation capabilities observed under the following conditions:
1 (i) Analysis of Packet Delivery Ratio (PDR): With a large number of sensors, efficiency in a PDR of SAB-UAS deteriorates.
2 (ii) Analysis of End-to-End (ETE) delay: There is a lesser delay compared to other methods. But with the increasing number of communication nodes, the delay in ETE is proportional.
3 (iii) Analysis of Throughput Transmission Rate (TTE): SABUAS has a better throughput rate compared to other authentication systems, and there are negligible deviations in TTE even when there were increased communication nodes.
4 (iv) Analysis of Routing Overhead (RTO): SAB-UAS seems to have tactful management of packet routing enhancing network performance and bandwidth usage.
The above simulations are essential