Design and Evaluation of Large-Scale IoT-Enabled Healthcare Architecture

Abstract

Employment of the Internet of Things (IoT) technology in the healthcare field can contribute to recruiting heterogeneous medical devices and creating smart cooperation between them. This cooperation leads to an increase in the efficiency of the entire medical system, thus accelerating the diagnosis and curing of patients, in general, and rescuing critical cases in particular. In this paper, a large-scale IoT-enabled healthcare architecture is proposed. To achieve a wide range of communication between healthcare devices, not only are Internet coverage tools utilized but also satellites and high-altitude platforms (HAPs). In addition, the clustering idea is applied in the proposed architecture to facilitate its management. Moreover, healthcare data are prioritized into several levels of importance. Finally, NS3 is used to measure the performance of the proposed IoT-enabled healthcare architecture. The performance metrics are delay, energy consumption, packet loss, coverage tool usage, throughput, percentage of served users, and percentage of each exchanged data type. The simulation results demonstrate that the proposed IoT-enabled healthcare architecture outperforms the traditional healthcare architecture.

1. Introduction

Recently, many people have neglected certain important aspects of their health, such as blood pressure, pulse rate, and other factors. In addition, some people frequently do not monitor their health and later discover health disasters. Moreover, some patients lack contact with their treating doctors due to difficulty communicating or long distances. Furthermore, some patients may need a specialized medical device that may not be available in the places surrounding them or their countries. These issues create an important challenge, which is to develop healthcare and use modern technology.

Currently, providing good healthcare to patients at a low cost is a lofty goal, but it is not sufficient. Patients who need urgent medical consultation and diagnosis as well as accurate follow-up, which may be available from a limited number of specialists, make sharing of experiences important. The sharing of medical information should occur at the level of the doctor and of the entire medical staff, such as nurses who assist the patient’s journey from illness to complete recovery through diagnosis and accurate follow-up processes.

The medical emergency specialty is considered one of the most important goals that must be integrated into a targeted healthcare system because patients need real-time diagnosis, treatment, and sensitive follow-up and have specific needs from specialists, which may sometimes be difficult. All objectives entrusted to the new healthcare system must be met with precision while ensuring the approved quality and safety standards. In addition to the reliability of scientific medical principles, due to the sensitivity of dealing with humans, the error rate should be the lowest possible, if not zero [1,2,3].

The Internet of Things (IoT) is described as a group of different devices connected to introduce additional services and improve others for citizens in various fields. Moreover, IoT technology differs from its counterparts in that it includes passive and active devices, which may help in joining billions of devices in our daily lives in various fields. There are many advantages of IoT technology, such as providing better service to customers, continuous technical improvement, and making real-time decisions. Therefore, it has been applied in several fields such as education, transportation, marketing, and environmental issues. For example, application of this technology in education has introduced new tools that are communicate with each other to raise the efficiency of the entire educational system. For the transportation field, self-driving cars, automatic traffic lights, road monitoring, parking assistance, and other examples became available in our lives. As regards marketing, IoT technology provides better services to customers and enhances business strategies. In addition, environmental issues such as detection of air and water pollution without human intervention has left a positive impact on people’s lives and prevented many disasters. Healthcare is one of the most important fields in which IoT technology has been applied. That is why healthcare companies have made numerous large investments in IoT-enabled healthcare. Recently, some communication forms support medical devices with the possibility of movement and presence in different locations, such as wearable devices, X-ray devices supported by Wi-Fi technology, and others. Moreover, IoT technology in healthcare has helped in accomplishing many tasks in various medical and pharmaceutical specialties, such as monitoring patients, determining treatment progress, offering shelter vaccines, monitoring drugs, and monitoring support equipment, such as cars and stores [4,5,6]. Figure 1 shows an overall view of IoT e-health system model.

On the other hand, scientists have realized that the development of devices used in human–machine interaction, such as cell phones, has reached its peak. This has resulted in researchers and many companies tending to use IoT technology to accomplish this type of communication. That is why the technology of IoT is associated with the field of brain–computer interface (BCI), which is considered as a new cornerstone of the interaction between the IoT objects and human. The BCI researchers have witnessed the transformation of human thoughts to physical actions, such as the IoT appliances and mind-controlled wheelchairs. BCI technology has many advantages, such as the invisibility of brain activities, which means that it cannot be repeated or noticed. In addition, the implementation of the reaction, which results from thinking, is achieved in a real-time manner [7]. Electroencephalography (EEG) is one of the important techniques that measures brain signals. The choice of signal-processing technology at each processing stage is very important for creating a robust BCI system. The description of different BCI processing techniques in addition to EEG classifications are stated in [8]. Nevertheless, the IoT-based BCI research field faces many challenges, such as measuring of certain brain activities affected by the state of emotions and some environmental factors. Furthermore, the technologies which are used to measure brain signals such as EGG have low signal/noise ratios in addition to lack of sufficient spatial resolution [9].

Based on the previous discussion, it is difficult to achieve all targets for IoT-enabled healthcare systems without a powerful communication system that relies on more than one method to cover devices and make them permanently connected without defect. Therefore, this paper proposes a large-scale IoT-enabled healthcare architecture in which more than one coverage tool is used to achieve communication between the medical and other devices regardless of their type or location. In addition, to obtain high efficiency in the proposed IoT-enabled healthcare architecture, it is supported with a simple and effective management scheme.

The contributions of this paper are as follows:

• the proposal of IoT-enabled healthcare architecture,
• applying a large-scale coverage scheme to guarantee communication between healthcare devices,
• usage of clustering and prioritization methodologies in the proposed IoT-enabled healthcare architecture,
• a simulated environment to measure the performance of the proposed IoT-enabled healthcare architecture, and
• discussion of the simulation results and recommendations.

The remainder of this paper is structured as follows. Section 2 discusses the literature review. Section 3 demonstrates the proposed IoT-enabled healthcare architecture. Section 4 presents the simulated environment specifications. Section 5 shows and discusses the simulation results. Finally, the paper is concluded, and future work is proposed in Section 6.

2. Related Work

Many researchers have provided healthcare systems based on IoT technology. The primary weakness of healthcare research relying on IoT is the lack of proper use of this technology, as it has been used in a very narrow range to connect a specific group of medical devices, such as sensors and wearable devices, and neglected many non-medical things such as those found in the transportation and security sectors, which are certainly used in the healthcare organizations. In addition, the communication between healthcare system nodes is based only on the Internet, although some necessary devices for healthcare systems may not have the ability to connect to the Internet. Internet signals sometimes may not be available in many isolated places.

Most related work is in the fields of monitoring applications, security applications, and hardware design. Many examples of related work are discussed as follows: Haghi et al. presented a platform to monitor performance indicators of health systems based on the idea of wearing them. This platform serves patients, paramedics, and the medical staff. The main weakness of this platform is in its evaluation [10]. Kumar et al. designed a healthcare system using Healthcare 4 and blockchain technologies. In addition, a comprehensive survey was conducted of all attempts to implement the blockchain in the healthcare system. The simulation infrastructure of this system has many limitations, such as the parameters, which are insufficient [11]. Min et al. introduced a security system to help users of IoT-enabled healthcare devices save their locations and usage. Moreover, it used the offloading rate to enhance computational performance and save energy [12]. Chanak et al. introduced an algorithm to solve congestion in IoT-based healthcare applications. This algorithm depends on the selection of routing paths depending on the data priority. In addition, a priority queue to enhance the reliability of the application was presented [13]. Aujla et al. proposed a double blockchain model to keep patient data safe in healthcare systems. In this model, the healthcare data are transformed from sensors to edge devices that send data to the cloud center using a tensor-based scheme [14]. Wu et al. demonstrated a design for a wearable patch sensor. The sensor measures many physiological metrics, such as photoplethysmography and blood pressure [15]. Bharathi et al. presented a technique called particle swarm optimization which was used to select cluster heads in IoT healthcare environments. Each cluster head is used to transmit health data to cloud devices, redirecting this data to fog devices [16]. Akkaş et al. proposed a WBAN-based architecture for biomedical applications. The system uses patient data, such as the oxygen ratio, pulse rate, and plethysmogram [17]. Ni et al. introduced a wearable device with low power consumption to detect the vibration of heart signals. This device helps the medical apparatus in healthcare institutions monitor critical patients, aiding in rapid intervention and saving lives [18]. Anand et al. introduced a device that can be attached to soldiers’ bodies to track their health and the current situation. The device uses IoT technology for data exchange between soldiers and control rooms [19]. Kadhim et al. presented a monitoring system for IoT-based healthcare organizations. This monitoring system permits doctors to check patients without physical interaction. In addition, a study on IoT-based applications in the medical field was introduced [20]. Misra et al. built a monitoring health system to avoid undesirable accidents. This monitoring system is based on IoT technology [21]. Campioni et al. proposed an algorithm for a radio frequency identification (RFID) network in which reader jobs are scheduled to save battery power. Hence, the network lifetime is increased [22]. Usak et al. presented a survey of healthcare research based on IoT technology. In this survey, the drawbacks and advantages of the proposed mechanisms are discussed as of 2018 [23]. Kumar et al. introduced an optimized technique for a virtual machine in cloud computing centers. This technique is used to decrease the execution time and increase the central processing unit utilization of healthcare devices and computing servers [24]. Jeyaraj et al. introduced an IoT-based application for physiological signal monitoring in the healthcare environment. A deep neural network is used to accurately predict the signal [25]. Asghari et al. proposed a monitoring system for patients in IoT-based medical environments. The monitoring scheme is used to gather predetermined data about the patients and predicts their future diseases [26]. Chui et al. proposed a monitoring system for patient behavior considering big data obstacles. In addition, they discuss the challenges, which are related to the security, privacy, and interoperability in the healthcare research field [27]. Kavitha et al. also proposed a real-time monitoring system for general health parameters using neural network methodologies [28].

3. Proposed IoT-Enabled Healthcare Architecture

The proposed IoT-enabled healthcare architecture consists of three components responsible for the three main functions: communication, management, and prioritization. The communicator component achieves communication between heterogeneous and homogenous healthcare devices, whether they are passive or active. The manager component accomplishes administration, adaptation, and management functions. The prioritizer component is responsible for classifying health data into several levels of importance. Each component and how the proposed healthcare architecture works are described below. Furthermore, a mathematical analysis is introduced. Moreover, for more clarification of the proposed healthcare architecture, a simple use case is demonstrated.

3.1. Communicator

Communication between health devices, regardless of type or number, is an important issue in the proposed healthcare system. There are three main communication challenges: coverage, device type (active or passive), and the numerous health devices. The first communication challenge is solved by using a powerful communication architecture that is described in [29]. The main idea of this communication architecture is to use three coverage tools. These tools are the Internet, high-altitude platforms (HAPs), and satellites. Each coverage tool is used in the case of the unavailability of the other two coverage tools. For example, if the device is not connected to the Internet and does not fall within the scope of coverage of an HAP, it can use the satellite as a coverage method. If there is more than one coverage method, the most-used method (Internet) is chosen.

Regarding the second communication challenge, different network technologies are used, such as wireless sensor networks (WSNs), RFID, mobile ad hoc (MANET), and cellular. Hence, RFID technology can be used for passive health devices to include them in the proposed healthcare system. The third communication challenge is solved by dividing the entire health environment into a number of clusters. Determination of each cluster size depends on the number of things (patients, users, devices, and others), the complexity of computation processes, which may be achieved within a period, in each cluster region, and the specifications of the cluster’s manager (see Figure 2). For example, assume that there is a healthcare unit in a densely populated area. Most of these residents in this area are elderly, which means that the rate of patients’ visits to this healthcare unit will be high, and the nature of the diseases of these residents may be classified as high risk. That is why this healthcare unit must be equipped with all of the equipment necessary to serve these types of patients. Hence, it is expected that the number of things and the size of exchanged data in this healthcare unit is extremely large and, of course, their rate of increase will be very high. Therefore, when integrating this healthcare unit into the proposed architecture, it is necessary to specify an appropriate number of homegrown managers (in case of considering this healthcare unit as one cluster) or dividing it into many simple clusters.

3.2. Manager

Management of the proposed healthcare system represents an important aspect due to the massive number of health devices communicating over a large scale. In addition, a massive number of gigabytes are gathered, sent, and received through the healthcare system.

The proposed healthcare system is divided into clusters, each comprising heterogeneous devices such as RFID-based, sensor-based, mobile, and other devices. Each cluster comprises a homegrown manager (HM). The number of devices in each HM region is determined depending on the size of each cluster and the activities executed within a short interval. Thus, a new HM may be created in the case of overloading on the current HM. Each HM monitors device activities, such as sending and receiving, and the type of transmitted data. The entire healthcare system is managed by the universal manager (UM). Hence, many communications should be achieved between the UM, HMs, computing servers, and devices. These communications are accomplished using control messages. These control messages are sent periodically. The time interval between each message transmission and the next one is determined depending on the message type (sudden event messages should be send regardless the bandwidth factor) and the current available bandwidth. The number of messages is directly proportional to the available bandwidth (i.e., if the available bandwidth is large, the number of management messages will be increased and the time interval between two consecutive messages will be decreased). The control messages transmitted through the healthcare system from UM to HMs, from HMs to UM, from HM to HM, from HMs to their devices in the clusters, and from devices to their HM are classified into four categories. The first category has three types of messages: guidance, stop, and inform. Guidance messages are used when the UM must determine the HM status, such as active or sleep. Stop messages are used to break the connection between some HMs and the healthcare system due to fewer devices in its cluster or insufficient energy to achieve its function (if the HM is an energy-based device). Inform messages are used to send regular information, which may be important or traditional for HMs. The second category has four types of messages: status, threshold, sudden, and prediction. Status messages are sent periodically to reflect the network status in the healthcare system (clusters), which comprises information such as the number of devices, size of transmitted data, and type of data most exchanged. Threshold messages are used when an element of the healthcare system reaches one of its indicators, such as the power or processing unit threshold, which means after a short period, it will be unable to perform its tasks, which leads to finding an alternative. Sudden messages are sent when an unexpected event occurs. Prediction messages are used when the HM has some signs of a future event, such as an increase in the number of devices or a huge data transmission in a special period. The third category has one message, called update. Update messages are sent from one HM to another when a device is transmitted from one cluster to its neighbor or when two HMs must cooperate to accomplish a special function, such as computing. The fourth category has one message, called new. New messages are used to inform the device that a new function is required or a new update has been implemented in the healthcare system. The fifth category has one message, called aggregate. Aggregate messages are used by HMs to gather information about devices under its management. The HM collects data about each device, such as the power consumption rate, number of sent and received bytes within a period, type of device, and other data. In addition, the HM can abstract these data into one result and send it to the UM.

Table 1. Management messages and their summarized functions.

Group #	Type	Description
First	Guidance	Used by UM to determine HM status (active/sleep/off).
	Stop	Used to break the connection between HM and its cluster.
	Inform	Used to send regular information to managers.
Second	Status	Used to describe the clusters status.
	Threshold	Used when an element of the healthcare system reaches one of its indicators (power or processing unit) threshold.
	Sudden	Used in case of an unexpected event occurrence.
	Prediction	Used when HM has some signs of a future event.
Third	Update	Used to update the cluster status.
Fourth	New messages	Used to inform the device that new function(s) is required.
Fifth	Aggregate	Used to gather information about devices in HM’s cluster.

summarizes the management messages and their description.

Based upon the above discussion, the number of messages increases such that it may represent an overload for the healthcare system from a communication perspective. Two strategies are executed to solve this problem. The first is that each group of messages is treated as one message with many flags. For example, the first category (i.e., from UM to HMs) has three message types: guidance, stop, and inform. Hence, one message comprises three flags in its header: a ‘G’ flag for guidance messages, an ‘S’ flag for stop messages, and an ‘I’ flag for inform messages. When the UM needs to send a special message type, its flag should be transformed into one, and the other flags stay at zero values. If two flags are settled into one value, two message types are sent in one message, and the data part in the message is divided into two main parts: one for the data of the first message and the other for the data of the second message. The first strategy can be executed on each message category; thus, the result is only four main control message types. Figure 3 contains a simple structure of one type of control message. The second strategy to decrease the control message overload is message prioritization (see Section 3.3).

The functions of HM are summarized as follows:

• Manage its cluster(s).
• Monitor the activities of each thing in its cluster(s).
• Update its cluster by deletion/addition of old/new thing(s).
• Predict the future events in its cluster(s).
• Control the message exchange process in its cluster(s).
• Gather information about the things in its cluster periodically.
• Send summarized report about its cluster to UM.
• Communicate with neighbor HMs.
• Select a computing center to process a specific data inside its cluster region.

The functions of UM are summarized as follows:

• Manage the entire domain of the proposed IoT-based healthcare environment.
• Select additional HMs for a specific cluster (If required).
• Achieve the load balancing issue between clusters.
• Achieve the fault tolerance issue inside clusters.
• Achieve the communication between clusters’ HMs.
• Evaluate the clusters in the healthcare environment periodically.
• Exchange between the messages’ transmission strategies.
• Select a computing center to process a specific data outside a cluster region.

3.3. Prioritizer

The IoT-enabled healthcare environment has numerous data that are exchanged between healthcare devices. In addition, IoT contains many heterogeneous networks, creating bottlenecks at any time. Therefore, processing all healthcare data at the same level of importance is difficult (if not impossible). Hence, a component that takes healthcare data as input and extracts these data for different performance should support the proposed healthcare architecture. This component is called a prioritizer. The data can differ from one health organization to another, making the level of importance different. Therefore, the prioritizer should consider this issue by providing n levels of importance. The value of n is changed from one health organization to another because the data importance is determined using a protocol by a group of specialists.

3.4. How the Proposed Healthcare Architecture Works

The proposed healthcare system comprises the direct related health devices and devices that may be needed for medical systems, such as cars. Hence, these devices are categorized into four main groups. The first group comprises sensing-based devices. The RFID-based devices (often passive), such as equipment, which are stored in the medical inventories are included in the second group. The third group comprises wearable devices related to health, such as blood pressure monitors. The fourth group includes cellular devices, such as smartphones and their applications. Using this categorization, Figure 4 illustrates a general view of the proposed healthcare system, including the numerous health devices of different types and locations. A WSN can connect devices in the first category. The second category of devices can communicate using an RFID network. The third device category can communicate using a MANET. A cellular network can connect the fourth category of devices. The four categories can be communicated using the Internet (or other alternative coverage tools), which leads us to deal with IoT technology.

One of the most important goals of the proposed IoT-enabled healthcare is to have a robust communication system in which health devices operate as if they were one device. Therefore, each device in the proposed healthcare system is covered using one of the three coverage tools: satellites, HAPs, and Internet. The priority order of coverage tools is the Internet, then HAPs, then satellites. This priority depends on the tools that are the most popular and easily available for each device. For constant location devices, it is simple to maintain coverage using the first selected tool, but for moving devices, such as ambulances or wearable devices, these tools may change when transferring from one location to another. Active devices, which have a processing unit, can easily connect to any coverage tool. For passive devices without a processing unit, RFID is used to achieve this type of communication [30]. The communication process can be accomplished in three statuses, device-to-device [31,32], device-to-computing center, and computing center-to-computing center. These communication statuses are accomplished over short (Internet or Bluetooth) and long ranges (Internet, HAPs, or satellites). Using different ranges ensures the healthcare devices are continuously connected to the healthcare system. The detailed description of the communication architecture is presented in [33].

The UM and HM monitor the healthcare system. The main idea of the UM and HM is based on monitoring and action. Monitoring is achieved by gathering data on predetermined parameters. Actions are taken based on the collected data after processing. Further, managers use control messages to execute their functions in the proposed healthcare system architecture. The communication system guarantees that multiple healthcare devices become one device regardless of the type or location, but the data transmission between devices should be organized and managed. Therefore, each device must send regular control data to determine its importance and available power. This information is determined by the management center using control messages. Hence, each device should maintain continuous communication with its HM. If the device moves from one region to another, the new HM detects this device in its region and sends a control message to that device and its old HM informing them of the new state of this device. The control message also informs the device of newly available computing centers (each with its priority). If a new device wants to join a healthcare system, it should send a join message to its HM asking for registration. The HM informs the device using another accept or reject control message. In addition, each HM knows whether any device leaves the system due to technical or communication problems by waiting a reasonable amount of time after a predetermined time in which the off-device should transmit its regular control message if the status of its cluster is not critical.

Moreover, the UM and HMs maintain the specifications (specs) of each healthcare device. The important fields in this specs file are the priority, current function, and current device parameters, such as power level. Additionally, the most important function of the HM and UM is determining a suitable computing center and its alternatives for each device (cloud, fog, mist, or edge). The computing strategy is selected for the device after considering three parameters: the device location, data prioritization, and data size. The device location is determined in order to select the nearest computing center. Data prioritization determines the sensitivity degree of the data required to be processed in the computing center. The data size is considered to determine which computing strategy can process this size of data. Most management processes are achieved between the HM and its devices. However, many management functions should be accomplished between the HM and UM. The UM selects an alternative for any failed HM. In addition, UM is considered a communication link between HMs. If any HM needs data about a specific device in such a cluster, it can query the UM. The UM can change HMs, assign new HMs, or delete HMs.

Elected specialists in the healthcare organization determine the data classification rules. These rules may change from one healthcare organization to another. The data are prioritized depending on a group of rules, such as the type of patients, specialties, medicines, and location. Health data are gathered through sink nodes that are assigned in the clusters and direct this data to the prioritizer. Then, the prioritizer applies the prioritization rules to the data and sends them to cloud computing centers. The cloud center computes each data point according to its importance level and available quality of service (QoS). Afterward, the cloud center responds to the user.

The output of the prioritizer is a number of queues equal to the number of data importance levels. First, the data are distributed among the priority queues depending on their importance. However, if the available QoS in such a queue is less than that required to process the data in that queue, the data importance is changed to the next level and transferred to a less important queue after decreasing its required QoS. The process continues until all healthcare data are serviced (in efficient networks), or some data with low importance are delayed until their required QoS is available. The UM should synchronize between healthcare organizations in the IoT environment regarding many issues, such as changing importance levels for specific data when changing the organization, calculating data importance levels using fog-computing centers instead of cloud-computing centers, balancing between computing center loads, and fault tolerance issues. The mathematical analysis of the prioritization work is stated in [34,35], in which the value of n equals five (see Figure 5).

3.5. Mathematical Analysis

The mathematical analysis stated in [36] is considered the base of this mathematical analysis. Suppose that the healthcare data arrives to the priority queues using Poisson distribution with rate equal to ${\mathsf{\lambda }}_x$, where ‘$x$’ is the data request index per unit time (second). The total number of requests equals ‘$T$’, so the overall arrival rate $\mathsf{\lambda }$ for all requests can be determined using Equation (1).

$$\mathsf{\lambda }={\displaystyle \sum }_{x=1}^T{\mathsf{\lambda }}_x$$

(1)

For simplicity, the data, which comes to the priority queues, is subject to the processing rule “first come—first served”. The prioritizer receives the data from the healthcare cluster with probability $P_1$ and the users receive the data after processing with probability $P_2$. The M/M/1 queue model is used for the system gateways with finite buffer. The M/M/C queue model is used for the computing centers. The processing time in each queue of the prioritizer is determined by 1/${\mu }_P$. The processing time of computing centers is determined by 1/${\mu }_C$, where $\mu$ is the service rate. For each private computing center, the M/M/C/K queue is used with processing time equal to 1/${\mu }_{C_i}$, where 1 < i < y and y in the number of computing centers. The probability of data staying in the prioritizer is determined using Equation (2), where $1-P$ is the probability that the healthcare data is forwarded to a prioritizer $P_j$ in a specific organization $O_i$ and $P^m$ is the probability that the data is served in a specific computing center where m is the total data in a specific node in the healthcare system.

$$O_i{P}_j({\pi }_m)=O_i{P}_j\left((1-P\right)P^m)$$

(2)

The mean of healthcare data in execution requests, $M{N}_{P_j}$, is determined using Equation (3). The number of healthcare data waiting requests for a specific prioritizer in a specific organization $D{N}_{P_j}$ is determined using Equation (4). The utilization of each prioritizer, $U_{P_j}$, is determined using Equation (5)

$$M{N}_{P_j}=O_i((1-P){\displaystyle \sum }_{m=1}^{\infty }m{P}^m)$$

(3)

$$D{N}_{P_j}=O_i{P}_j({\displaystyle \sum }_{m=1}^{\infty }m{\pi }_{m+1})$$

(4)

$$U_{P_j}=O_i{P}_j(1-{\pi }_0)$$

(5)

The mean response time, $M{R}_{P_j}$, is determined using Equation (6), and the mean waiting time, $M{W}_{P_j}$, is determined using Equation (7).

$$M{R}_{P_j}=\frac{O_i((1-P){{\displaystyle \sum }}_{m=1}^{\infty }m{P}^m)}{\mathsf{\lambda }}$$

(6)

$$M{W}_{P_j}=\frac{O_i{P}_j({{\displaystyle \sum }}_{m=1}^{\infty }m{\pi }_{m+1})}{\mathsf{\lambda }}$$

(7)

The number of satellites and HAPs that are required to cover the entire earth is determined using Equations (8) and (9), where $r_{earth}$ is the earth radius, $E$ is the elevation angle, and $h$ is the height of antenna.

$$N_{SH}=\lceil \frac{4\ast \pi }{\left(1-\cos \left(\theta \right)\right)\ast 3\sqrt{3}}\rceil$$

(8)

$$\theta =\left[{\cos }^{-1}\ast \left(\frac{r_{earth}\ast \cos \left(E\right)}{r_{earth}+h}\right)\right]-\left(E\right)$$

(9)

3.6. Simple Use Case

Many patients with diseases such as Alzheimer’s, diabetes, weakness, mental disability, and addiction may have unpredictable reactions. Sometimes, these types of diseases make the patient unbalanced and lose consciousness or awareness. In addition, most of these diseases have become widespread all over the world, and the most vulnerable to these diseases are elderly. Moreover, there are many accidents that have occurred for these types of patients due to leaving them without monitoring or tracking. Having permanent monitoring by a person is a very difficult process, especially since these patients may not have the money or qualified persons for this in addition to the safety factor. Moreover, placing monitoring cameras in their locations has its drawbacks because anyone who feels that he is being monitored may have bad feelings that hinder or delay the duration of his treatment. In addition, this type of monitoring requires that the patient should have limited movements, which is not desirable. This is why applying IoT technology in the healthcare field will help these types of patients.

Based upon the above discussion, many technologies are needed to achieve the monitoring target. For example, the clothes of these types of patients can be supported with RFID through which we can track them. In addition, by strengthening their bodies with nanosensors, we can measure general health parameters such as blood pressure, pulse, and degree of awareness. As stated above, these patients are spread all over the world, including in poor and remote places, which means that the Internet service may be unavailable and the communication with them will be so difficult (if not impossible). To face this challenge and guarantee the communication/integration of patients to/in the proposed architecture, HAP and satellite are used. Hence, if there is a relapse for any patient, their data will be sent to the predetermined health organization in a real-time manner. In addition, a suitable ambulance, which is closest to the patient and equipped with all of the medical requirements and teams that are appropriate to the medical status, will be called and moved towards the patient’s current location. In addition, the patient’s health parameters are measured and sent to their treating doctor(s) and emergency specialists (if required). Furthermore, if the treating physician needs expert advice, full patient data can be made available to the selected expert in a real-time manner. Moreover, the initial treatment plan for the patient (such as the required medication, blood transfusion, preparing the surgical room, intensive care) will be constructed using artificial intelligence (AI) healthcare applications [37]. Finally, the patient’s location will be accurately determined and all of their current specifications can be automatically sent to their relatives (see

Table 2. Needs for the use case construction.

Tool	Mission	Network	Coverage
Medical Sensors	Monitoring	WSN	Internet	HAP	Satellite
RFID	Tracking	RFID
Mobile	Connection	Cellular
Health Software	Decisions	AI
Specialist Temporary Cooperation	Discussion	MANET

).

4. Simulation Infrastructure

Simulating of IoT environments and making the simulated IoT environment work as a large multilateral health organization are two main objectives in our simulation process. The first objective is achieved using the architecture proposed in [38,39,40]. In this architecture, the satellite, HAP, and Internet are simulated and cover many devices at many locations. Using Internet coverage tool has the first priority due to its wide spread. Then, HAP will be used in the event that Internet is unable to cover the healthcare things. The satellite coverage tool will be used in the case where both Internet and HAP are unable to achieve their coverage mission. Determination of the number of satellites and HAPs depends on two main factors, the size of the coverage area and cost. In this simulation, one satellite and five HAPs are used to cover the healthcare things. The coverage tools and things are arranged into five layers, satellite/HAP/HAP/Internet/things. The things in the fifth layer can be transformed to be in the second layer (in case of using satellite), or the third/fourth layers (in case of using HAPs). The type of satellite is LEO, and its specifications are altitude, link bandwidth, inclination, and delay, and the corresponding values of satellite specifications are 800 km, 25 mbps, 86 degree, and 7.8 ms, respectively. The HAP specifications are radius, altitude, noise, and height of antenna, and the corresponding values of HAP specifications are 50 km, 28 km, 5 dB, and 22 km, respectively. In addition, the simulation description of WSN, RFID, MANET, and cellular networks is stated in this simulated architecture. The WSN specifications are RF power, covered area, and data rate, and the corresponding values of WSN specifications are −10 dBm, 1200 km × 1200 km, and 250 Kbps, respectively. The RFID specifications are data rate, sensing range, power, and frequency average, and the corresponding values of RFID specifications are 2 Mbps, 5.4 m, 145 dBm, and 915 mHz, respectively. The MANET specifications are packet size, area, and number of requests, and the corresponding values of MANET specifications are 1 Mb, 500 m × 500 m, and 3750, respectively. The cellular network specifications are signal to noise ratio (SIR), BS transmission power, and sensitivity of the stations. The corresponding values of cellular network specifications are −9 dB, 17 dBm, and −65 dBm. There are 2000 routers distributed over 10 countries. Each router can serve a range of nodes from 300 to 500. The devices can be directly connected to the routers or by using sink nodes. Figure 6 shows a simple view of simulation architecture.

The second objective of the simulation is to make the networks forming the IoT enjoinment connect devices with a direct relationship to the medical field and devices serving the medical system in general. In this simulation, the devices to be included have been studied so that the simulation system can include any device with a medical relationship under the umbrella of an IoT-enabled healthcare system. Initially, devices were divided into “active” and “passive”. As stated above, active devices can connect to any coverage tool, such as the Internet, and have a processing unit (optional), and passive devices cannot contact any coverage tool. For this, active devices are communicated using a WSN, MANET, or cellular network, and passive devices are communicated to the IoT-enabled healthcare system using RFID networks supported with other tools, such as RFID tags (see Figure 7).

Table 3. Partial of simulation parameters.

Parameter	Value
Number of patients	50,000
Number of organizations	20
Number of users	100,000
Number of active devices	200,000
Number of passive devices	300,000
Distance between organizations	Random range (200–1000 km)
Number of wearable devices	30,000
Number of body sensors	15,000
Number of mobiles with medical applications	10,000
Number of HAPs	5
Number of satellites	1
Distance between users	Random in range (10–1000 km)
Distance between patients	Random in range (10–100 km)
Computing centers	Cloud and fog
Data types	Text, image, and multimedia
Transmission channels	Full duplex
Types of sensors	Heterogeneous
Number of medical devices	1000 (distributed)
Number of monitoring devices	20,000
Number of importance queues	5

lists the values of the medical variables required to represent the healthcare system in the IoT environment.

5. Results and Discussion

The proposed architecture simulation results are compared to those of the traditional healthcare architecture. This traditional architecture has only one coverage tool (the Internet). In addition, it processes the healthcare data as being on importance level. Moreover, the cluster idea is missed, and the management functions are neglected. The performance metrics to measure the efficiency of the proposed healthcare architecture are the delay, energy consumption, packet loss, coverage tool usage, throughput, number of serviced users, and percentage of each data type exchanged through the two architectures (IoT-enabled and traditional healthcare).

Figure 8 displays the average delay for the two healthcare architectures. The delay performance metric is measured by calculating the average of processing, transmission, buffering, and propagation delays for sent packets until they reached their destinations. The x-axis represents the time of simulation, and the y-axis represents the average delay in seconds. The overall average delay for the proposed IoT-enabled healthcare architecture is less than that of the traditional one because the proposed architecture comprises more than one coverage tool. Therefore, each user can easily change its coverage tool, which means that the healthcare data is exchanged through powerful transmission channels (high bandwidth). The hesitations, which appear in the plots of both healthcare architectures, result from the changing of coverage tools in addition to the dynamic changes in joining/leaving things in the architectures over the simulation time.

Figure 9 depicts the energy consumption rates for the two healthcare architectures. The energy, which is consumed by sensing, communication, and processing functions, is used to calculate the energy consumption average for the energy-based nodes. The x- and y-axes represent the simulation time and average energy consumption rates, respectively. The value of energy consumption for the proposed IoT-enabled healthcare architecture is less than that of the traditional healthcare architecture. This result is explained by the availability of alternative coverage tools that allow all of the healthcare devices to communicate in the IoT-enabled healthcare architecture. Thus, the transmitted data are distributed among a greater number of devices than the traditional architecture. Therefore, the size of the transmitted data of each device is decreased, leading to lower energy consumption rates. Furthermore, decreasing the size of processed data, which is achieved by the prioritizer component, is an important factor that positively affect the energy consumption rates. The energy consumption rates are relatively increased over the simulation time for both healthcare architectures due to the increase in data processing requests.

Figure 10 presents the packet loss ratio for the two healthcare architectures. The x- and y-axes represent the simulation time (minutes) and packet loss ratios over the simulation time, respectively. The packet loss percentage for the proposed IoT-enabled healthcare architecture is less than that of the traditional architecture because the congestion rate increases in the traditional healthcare architecture due to using only Internet as a coverage tool, increasing the packet loss ratio. Moreover, for the traditional healthcare architecture, increasing of transmission and processing data leads to increasing of packet loss ratio. The small hesitations, which are found in the two plots, are explained by sudden changing in the available bandwidth values due to bottleneck occurrence.

Figure 11 displays the employment of coverage tools. It is measured by the percentage of users that have benefited from each coverage tool as first, second, or third priority within a simulation time. This performance metric is selected to ensure that each device can easily exchange between different coverage tools in the proposed IoT-enabled healthcare architecture. The x- and y-axes represent the coverage tool (first, second, and third priority) and usage percentages, respectively. The Internet coverage tool has the largest values. The HAP coverage tool follows the Internet, and the satellite coverage tool has the least usage percentage. This is because the Internet has first priority to cover things in the proposed IoT-enabled healthcare architecture. In addition, in some situations the satellite and HAP are used for the things that loss the Internet signal, which means that both of them may have first or second priority (according to their coverage areas).

Figure 12 illustrates the average throughput for the two healthcare architectures. The x- and y-axes represent the simulation time and average throughput divided by 10E6, respectively. The average throughput for the proposed IoT-enabled healthcare architecture is greater than that of the traditional architecture over the simulation time. This is explained for the traditional healthcare architecture by increasing packet loss and delay values causing a negative effect for the size of data that is correctly received and a decrease in the average throughput. The fluctuations in the two plots come from the network bottlenecks and numerous dynamic joining and leaving in the architectures that cause sudden changes in the size of the transmitted data.

Figure 13 presents the percentage of served users in the two healthcare architectures. The x- and y-axes represent the number of users per simulation time and the servicing percentage, respectively. The red points represent the proposed IoT-enabled healthcare architecture and are between approximately 90% and 100%, and the black points represent the traditional healthcare architecture and are between approximately 80% and 91%. Therefore, the servicing percentages over the simulation time for the proposed IoT-enabled healthcare architecture are larger than those of the traditional healthcare architecture. This outcome is because of the availability of multiple coverage tools, such as satellites and HAPs, in addition to Internet. In addition, the selection flexibility of computing resources, which are connected to the proposed IoT-enabled healthcare architecture, increases the number of benefited users.

Most of healthcare things, such as doctors, patients, managers, employees, pharmacists, and the vast majority of medical and non-medical devices, use text data to complete the cooperation between each other. In addition, many medical devices use other types of data such as multimedia (video and audio) in addition to images. Hence, this performance metric is measured to ensure that the proposed IoT-enabled healthcare architecture can transmit the different data types required by medical fields to achieve its functions. Figure 14 depicts the percentage of each exchanged data type through the proposed IoT-enabled healthcare architecture. The text data type has the highest percentage, followed by the image data type. Finally, the multimedia data type has the least percentage value.

6. Conclusions

In this paper, a large-scale IoT-enabled healthcare architecture is proposed. This architecture consists of three main components: a communicator, manager, and prioritizer. To ensure that the maximum number of medical devices is covered, three coverage tools are used under predetermined priority rules. These coverage tools are satellites, HAPs, and Internet. The control of the coverage tools is the function of the communicator. For easy management objectives, the manager is used to transform the proposed architecture into small clusters. The prioritizer is used to create several importance levels for the healthcare data. A network simulation package (NS3) is used to measure the performance of the proposed IoT-enabled healthcare architecture. The simulation results demonstrate that the proposed IoT-enabled healthcare architecture enhances the performance of the traditional healthcare architecture. The delay decreased by 19.211%, and the energy consumption decreased by 11.357%. The packet loss decreased by 26.886%, whereas the throughput increased by 22.999%. The percentage of served users increased by ≈10%. The usage of coverage tools and the percentages of each exchanged data type were measured. Finally, the future work of this research is summarized in executing of more intensive simulation experiments until find the optimal utilization of coverage tools while maintaining a balance between efficiency and cost factors. Moreover, trying to apply other methodologies to decrease the size of the data that is exchanged through the proposed IoT-enabled healthcare architecture.