Here, we describe various types of remote medical services and define common elements and structures that serve as a basis for remote medical services in the IoT environment. In addition, the structural and functional limitations of the remote medical environment based on the defined elements are extracted.

First, 'remote treatment' is one type of medical service that is being developed that strives to overcome geographic and time-related obstacles. Remote treatment service was introduced for prisoners of the Full Sutton Prison located in a suburb of the city of York by Airedale NHS Foundation Trust of the United Kingdom in 2006. The kiosk-type unmanned telemedicine system 'HealthSpot Station', which was introduced on Consumer Electronics Show (CES) in 2013, is an installation-type clinic where various types of physical examination in addition to video treatment by a doctor are possible within the designated space. This includes many services, such as WellDoc BlueStar, which is a remote diabetes management mobile software approved by the US Food and Drug Administration, and PhoneDOCTORx, which allows a professional medical team in a remote location to view the abnormal conditions that occur during regular nursing care of a facility resident via video phone to assist in decision-making [56].

Next, 'remote monitoring' started with technology introduced in the aerospace program of the former Soviet Union to detected the signals of living organisms for remote communication. This technology also made it possible for NASA to collect biometric data and check predictable physical funct ions in advance to allow for remote response while astronauts conducted their work. According to the Frost & Sullivan survey, the growth rate of the remote monitoring market has maintained double digits in the past 10 years, and it is currently expected to achieve remarkable advances in the development of wearable devices and data sensing technology centering on IoT. Ultimately, this research will enable medical consumers, such as cancer patients and patients with chronic illnesses, to have their health condition constantly monitored by medical staff and to have a 'designated doctor' who can provide them with medical services regardless of time and place [56].

Finally, Remote 'Diagnosis·Treatment·Control' management are being attempted along with various technological developments. First, da Vinci (Intuitive Surgical Inc., Sunnyvale, CA, USA), which is famous for application in robot-assisted surgical procedures, has proven the effect by attempting an operation from a remote location focusing on many large domestic hospitals. In addition, IBM Watson analyzes big data within 3 seconds, focusing on 600,000 clinical data and 42 medically related DB, in providing a practical advice service to medical staff. Furthermore, Asklepios Hospital, located in Hamburg, Germany, has attempted to utilize augmented reality by overlapping an actual photograph based on a pre-operation clinical image of the patient in operation. Various efforts are being made around the world to provide diverse remote medical services, including Changi General Hospital in Singapore, which uses QR codes to manage personalized administration for each patient via robot [56].

The technology that serves as the basis for the remote medical services mentioned above is presented in Figure 1. The structure makes data collection from sensors and devices, pro cessing, transmission, execution, analysis, and other actions possible to support remote medical services. For remote medical services in such a structure, change in service is inevitable in accordance with sub-elements, including network technology, protocol, sensor, device, etc. Also, there is a constraint in that the structure should be restructured whenever a new service is created. The same problem occurs when a security service and technology is applied to a remote medical environment.

When the required factors of IoT, human, object, and service are applied to the remote medical environment, 'human' is subdivided into classifications of patient, medical personnel, and non-patient. 'Object' refers to all objects and infrastructure on the internet located around the 'human' element, in other words patients, medical personnel, and non-patients. Lastly, 'service' is media that expresses and supports medical communication between human-object-infrastructure. Accordingly, remote medical services with an IoT architecture can be classified into the following three layers: 1) the perception layer, which receives health signal entries from various sensors; 2) the network layer, which takes on the role of transmitting the health data from the perception layer to the application layer; and 3) the application layer, which provides various services and interfaces via the health data received from the network layer [789].

Consequently, the security elements of the remote medical architecture should be included at the initial design level because simple cyber-attacks can be directly connected to threats against patients' lives. A service-oriented architecture is flexible and can support a combination of applications depending on telemedicine environmental changes. Therefore, this study was focused on designing a service-oriented architecture and security for remote medical services.

If remote medical services are carried out in the IoT environment, security vulnerability in the general remote medical setting and security vulnerability in the IoT environment will both be present. Therefore, in this section, the remote medical security threats in the IoT environment that take both conditions into consideration are extracted, and then security demands regarding these threats are described.

The proposed framework is composed of a total of five layers, including the Application Service Layer, Service Support Layer, Network Layer, Perception Support Layer, and Perception Layer. Figure 2 shows the structure of the proposed framework.

The Application Service Layer offer a convenient user interface to patients, medical staff, and non-patients. It is supplied with data and resources from the Service Support Layer based on requests by the service. In each of the service applications, independent service with reinforced security exists.

The Service Support layer manages data and resources which serve as the basis for the efficient performance of the Application Service Layer. It takes the data transmitted by the Perception Support Layer and stores it in an integrated database through the second access controller. The Service Mgr (Service Management), Data Mgr (Data Management), Resource Mgr (Resource Management), and System Clock manage and supply service, data, resource, and time data regarding the application service. The A_En/Decryptor encrypts (or decrypts) the data received from the A_Communicator. The A_Communicator receives the data from the P_Communicator, and then the data is forwarded to the second access controller (or A_En/Decryptor).

The Network Layer functions as a transmission medium and protocol. It handles direct and indirect data communication between the Service Support Layer and Perception Support Layer.

The Perception Support Layer collects the data received from the Perception Layer and controls bad data and unauthorized access. After that, it performs pretreatment to ensure that the data collected from different types of devices and sensors can be used appropriately. The P_Encryptor encrypts the data received from the Classifier, and the P_Communicator sends the data that was received from the P_Encryptor and Classifier.

The Perception Layer is the domain in which the device (or sensor) that creates data exists. Various types of data created in this layer are transmitted to the Perception Support Layer.

In the next section, we focus on the perception side layer and application side layers, which are the core territories of remote medical services with IoT, in adding each network elements to propose framework, which will be separated into two sections, namely, the A-N section and N-P section, in the explanation. The N-P section refers to the Perception Support Layer, Perception Layer, and Network Layer, while the A-N section refers to the Application Service Layer, Service Support Layer, and Network Layer.

Here, we present a medical service scenario in which a secure remote medical service is provided in the IoT environment with proposed framework application. Table 7 defines the medical service-oriented terms and Figure 5 displays a diagram based on the proposed framework. Whether security technology is applied to each of stages ①-⑥ is explained in relation to the process.

① Transfer of perception data: verify through DA between device (or sensor)-device (or sensor), device (or sensor)-collection gateway, and transmit data securely through SC, DED during data transmission.

② The collected data transferring and preprocessing from device (or sensor): After completing DA on the collection gateway, CBAC and RBAC should be applied to transmit the collected data. If additional security is required, DED and SC should be used because they make it possible to provide strong security service.

③,④ Transferred data update and access to electronic medical records: The UA used when accessing the electronic medical records of the medical staff and others, including DED, SC, CBAC, and RBAC are shown in ①.

⑤,⑥ Providing the services to users: DA is used when device and sensor transmit data from the user (non-patient, patient). On the other hand, when the user wishes to access a service, the UA is used. The application of CBAC and RBAC is required during data transmission between user-service; SC and DED are selectively provided for stronger security.

Excluding survey-type theses from the previous research considered, four studies that took confidentiality, integrity, availability, and privacy into consideration were used in conducting a comparative analysis against the proposed framework. The comparative analysis of the proposed framework was conducted with the security element as the standard, which was the selection standard of the previous research and previous studies. The security comparison results are shown in Table 8.

• Confidentiality: The proposed framework provides confidentiality via device- and user-based authentication and status- and role-based access control and security channel, encryption, etc., for the purpose of confidentiality between device (or sensor), device (or sensor)-collection gateway, service and user. Zaidan et al. [14] mentioned the network structure and secure transmission regarding health information exchange via verified encryption and hash algorithm. Li et al. [15] mainly focused on secure patient-oriented personal health record access and efficient key management. Savola et al. [16] suggested maintaining confidentiality via an adaptive management model and focused on the role of security metrics. Moosavi et al. [17] tested authentication and authorization in the IoT healthcare environment through a certificate-based DTLS handshake protocol-based structure and proposed a technology for IP confidentiality in the IoT environment.

• Integrity: If an integrity verification measure is not available during data transmission and reception, there is a concern about too much data modulation; this is particularly important in the case of medical data. Accordingly, to guarantee integrity, it should be possible to check for data modulation in each section. In the proposed structure, the integrity of transmitted and received data in each section is guaranteed through the first and second authentication. Also data integrity can be maintained by the cryptographic algorithm. Zaidan et al. [14] proposed an encryption and hash algorithm for network-focused health information exchange; verification is not performed in each section. In addition, as there is no mention of non-repudiation, it is difficult to ensure accountability regarding medical care. Li et al. [15] attempted to prevent access to unauthorized self-health information and to ensure accountability regarding data access. In Savola et al. [16], role metrics on access control exists; however, there is no specific mention about that method of access control. Moosavi et al. [17] consider the only integrity of data in transmit-receive section.

• Availability: The framework proposed in this paper makes it possible to selectively apply encryption and a security channel that excludes authentication in each section, in providing confidentiality and integrity; thus, flexibly securing availability is possible. The framework proposed by Zaidan et al. [14] was designed only taking confidentiality and integrity into consideration. The method proposed by Li et al. [15] raises a concern because it can decrease availability in an environment where large amounts of data are obtained from various types of devices and sensors. Savola et al. [16] proposed availability metrics that include various elements, such as QoS, resilience, scalability, etc., but there is no specific mention about details. The structure proposed by Moosavi et al. [17] is structurally dispersed in key management and is comparatively more secure than the state of the art centralized delegation-based structure; therefore, it is more secure from availability attack.

• Privacy: The proposed framework decreases the exposure of sensitive information compared to other studies by applying a security channel and encryption during the transmission of sensitive information between network sections. Also, privacy is made more secure by only allowing personnel with authorized roles to access data within the permitted time regarding specific medical staff and by applying UA, CBAC, RBAC during EMR access by the medical staff. The frameworks proposed by Moosavi et al. [17] and Zaidan et al. [14] do not mention any technology to provide privacy. The method suggested by Li et al. [15] allows flexible handling of personal health record access policy. Savola et al. [16] discussed privacy protection through privacy metrics, but they did not suggest a specific method.

Utilizing the databases of PubMed, Springer, IEEE, Science-Direct, BMC thesis, 'remote medical' and 'IoT' were searched under the AND condition, and 'telehealth', 'mobile health', 'telemedicine' were each searched under a combination of OR condition. The search range was 2011 to 2015. As a result, approximately 1,200 papers were searched, and among these, a total of 30 were extracted based on a standard of abstract containing security elements. As with the security threats classified in Table 1, the searched papers were classified into device (or sensor), infrastructure, and service for discussion of the previous research in each category.

• Device (included sensor): Doukas et al. [10] described digital certificates and PKI data encryption based on a gateway for aggregation of health sensor data and to resolve security problems. Hsu and Pan [11] proposed agent-based telemedicine based on a P2P networking architecture. Camara et al. [13] explained the main security goals for the next generation of implantable medical devices and analyzed the most relevant protection mechanisms. Simplicio et al. [18] presented a lightweight framework for security. It focuses on protection of shared data's collection and lost/stolen device's data.

• Infrastructure: Saleem et al. [19] propose a framework for security based on IEEE 802.15.4. Also, in [12], the security vulnerabilities and major attacks in the context of wireless body area network (WBAN) were identified. Zhang and Zhang [20] introduced a secure and flexible platform based on IoT and cloud computing that uses short-distant ambient communication protocols for medical purposes. Shini et al. [21] described data storage and sharing through cloud computing. Their study highlights the different types of security problems that affect cloud users. Chen et al. [22] proposed an enhanced authentication scheme that overcomes the weaknesses inherent in Khan et al.'s scheme, and they demonstrated that their scheme is more secure and robust for use in a telecare medical information system. Al Ameen et al. [12] described security and privacy issues in WBAN, and they proposed measures for healthcare application in WBAN. Li et al. [15] focused on a multiple data owner scenario and divided the users in the PHR system into multiple security domains. Doing this greatly reduces the key management complexity for owners and users. Jiang et al. [23] proposed a secure and efficient authentication scheme with user privacy preservation which is practical for a telecare medical information system. Das and Goswami [24] proposed a robust remote user authentication scheme for connected health care that preserves uniqueness and anonymity. Kim and Lee [25] proposed cryptanalysis that discourages any use of the two schemes under investigation in practice, and they revealed some subtleties and challenges in designing this type of scheme. Das and Goswami [26] proposed a novel and secure biometric-based remote user authentication scheme to withstand the security flaw found in Awasthi-Srivastava scheme and enhanced the features required for an idle user authentication scheme. The architecture proposed by Moosavi et al. [17] is more secure than a state-of-the-art centralized delegation-based architecture because it uses a more secure key management scheme between sensor nodes and the smart gateway. Zaidan et al. [26] proposed their secure framework for health information transmission within a central cloud-based model.

• Service for remote medical: Savola et al. [16] proposed adaptive security management mechanism based on security metrics. And they described protection methods of confidentiality, integrity, availability in eHealth IoT application. Shin [27] introduced a framework for secure remote health-monitoring systems using a realistic risk model for sensor-data quality. Abie and Balasingham [29] described a risk-based adaptive security framework for IoTs in eHealth that estimates and predicts risk damages and future benefits using game theory and context-awareness techniques. The framework proposed by Al-Haj and Amer [29] is implemented on a block-level of the partitioned-image for integrity, thus enabling the localized detection of tampered regions.

In Table 9, the searched papers were first classified into device, infrastructure, and service. Then, they were sorted according to how they include the main security elements, such as confidentiality, integrity, availability, and privacy in understanding the characteristics and tendency of previous researches related to remote medical security.

In the infrastructure field, it is evident that many studies took various security elements into consideration. However, it is evident that there has been less research in this area than in the device and service fields. Accordingly, there is a need to reflect various security elements in all domains of service, infrastructure, and device.

As the demands and expectations regarding medical services are increasing, various new medical services are being introduced. In particular, there is much interest regarding possible medical activities via remote medical services from long distance, and this is being combined with the new concept of the IoT. However, security element support has remained weak, with security being impossible in some cases or only partial security elements being provided regarding the modification of existing services and the creation of new service. Accordingly, remote medical security in the IoT environment requires a new structure and method to prevent various threats directly connected to the safety of patients' lives.

This paper proposed a service-oriented security framework for remote medical services in the IoT environment. The proposed framework is a service-oriented structure that can support dynamic security elements in accordance with demands regarding various types of new remote medical services that will be created in the IoT environment, which will soon be realized. It enables the provision of secure services that guarantee confidentiality, integrity, and availability for all, including patients, non-patients, and medical staff. Security elements are realized via role- and situation-based access control, user access control, data encryption, and the provision of a security channel.

In future studies, we will conduct an in-depth study regarding techniques that can satisfy demands in accordance with key generation and management. In addition, we plan to enhance the efficiency and availability of the authentication method in the IoT remote medical environment.