2020 7th International Conference on Internet of Things: Systems, Management and Security (IOTSMS) | 978-0-7381-2460-5/20/$31.00 ©2020 IEEE | DOI: 10.1109/IOTSMS52051.2020.9340231 Security and Privacy of Medical Internet of Things Devices for Smart Homes Paige Harvey1 , Otily Toutsop1 , Kevin Kornegay1 , Excel Alale2 , and Don Reaves3 , 1 Department of Electrical and Computer Engineering, Morgan State University, Baltimore, MD 21251 USA Department of Electrical and Computer Engineering, University of Maryland, College Park, MD 20742 USA 3 Frederick Douglas High School, 2301 Gwynns Falls Pkwy, Baltimore, MD 21217 pahar9@morgan.edu, ottou1@morgan.edu, kevin.kornegay@morgan.edu, ealale@terpmail.umd.edu, doreaves@gmail.com 2 Abstract: The Internet of Things, IoT, provides various applications in the health and care sector, enhancing assisted living, and the ability to remotely monitor patients in hospitals, at homes, and in care facilities [1]. Additionally, many patients who require constant health monitoring prefer the comfort of home monitoring to hospital environments [2]. Therefore, with the increased interest in smart home configurations and the need for Medical IoT, MIoT, device development, we may witness a rise in in-home smart healthcare set-ups. In this paper, we present a Ph.D. research proposing an improve architecture for ensuring security and privacy of smart medical devices in smart home environments. Index Terms—Home Automation, Healthcare, Internet of Things, Internet of Medical Things, Medical IoT, Privacy, Smart Homes, Security I. T HE C ORE P ROBLEM OF P H D R ESEARCH S the adoption of IoT expands, two emerging areas that have gained recent attraction are smart homes and healthcare. The interconnectivity of smart medical technologies in smart home environments may present several benefits, including Remote Patient Monitoring, RPM, or telehealth, which allows patients to use smart medical technologies along with mobile devices to capture vitals, real-time, before getting transmitted to medical healthcare professionals, HCP, for further evaluation, and overall healthcare cost reduction through limiting hospital visits for every medical-related inquiry. Unfortunately, the increasing number of smart home devices installed on home networks makes them an attractive target for hackers [1].Additionally, not every home automation or MIoT, also known as the Internet of Medical Things, IoMT, device owner adheres to proper internet and networking etiquette when installing appliances in the home. Therefore, several of the risks that must be assumed by users, as well as manufacturers, include: malicious actors discovering the brand and dosages of medications prescribed, the time patients might leave and arrive home from work, which devices do patients rely on for medical purposes, and the email accounts that are associated with the sensitive data, just to name a few. However, the biggest challenge is the interoperability of devices, which can lead to a network being exposed to new security vulnerabilities and additional risk [3]. Therefore, throughout this research, we seek to evaluate heterogeneous devices, manufacturers, and communication protocols to determine a capable system architecture that will allow users to maintain the integrity of their device’s behavior and the optimal security and privacy levels associated with patient data. In [4], the STRIDE method is utilized to provide security and privacy recommendations for various IoT applications, in- A cluding smart homes and healthcare. The acronym for STRIDE can be found below: • • • • • • Spoofing – a potential attacker pretending to be the user of the application or system Tampering – modification of systems and/or resources that are not supposed to be modified Repudiation – claiming that nothing will interfere with the system, regardless of whether there is an interference present Information Disclosure – exposing information to people who do not have the authorization to view that information Denial of Service – attacks that deter a system from providing services by causing it to crash or significantly degrade performance. Elevated Privileges – malicious users increasing their privileges on a system Using STRIDE as a basis, Microsoft’s threat modeling approach was also leveraged to identify perceived threats as well as provided further recommendations for ensuring security and privacy concerns are addressed for various scenarios. In essence, a threat model is a system that recognizes, classifies, organizes, and rates threats based on a thorough analysis of an organization’s architecture, helps execute an organized way to deal with security business impact, and aids in describing the attack surface of edge devices. The threats that have been identified for both smart homes and healthcare include: data collection and control issues, hacking of smart home devices to gain access to the home and resources, and manipulation or hacking of devices to obtain health records or disrupt service. Ultimately, the authors of this work suggest a security and privacy by design approach for IoT device development and routine monitoring of networks for suspicious behavior as well as segregation of user-level and administrative privileges in smart homes. In the following paragraphs, we will discuss related work. In [5], the authors Authorized licensed use limited to: Carleton University. Downloaded on June 19,2021 at 21:51:22 UTC from IEEE Xplore. Restrictions apply. address privacy awareness through proposing a system model that centers around assessing user preferences for data collection, storage, analysis, and delivery, resulting in the subsequent anonymization and obfuscation of specified patient-sensitive parameters before being sent to designated institutions. In [2], the authors explore a smart intensive care unit, ICU, scenario to propose a system architecture that reduces the cost associated with healthcare and improves real-time patient monitoring using sensors. Leveraging ICU bedside monitors, Microsoft’s XBOX KinectTM sensors, for monitoring patient vitals and movement, and an additional sensor board to detect environmental factors, gathering information about the indoor environment, were evaluated. With the implementation of various servers and databases, a user-friendly interface was developed to allow doctors and nurses to stay informed with and implement preventative measures when necessary. In [6], the authors present a system architecture that addresses data acquisition and transmission, cloud processing, and analytics for RPM and healthcare cost reduction. Additionally, in this approach, machine learning techniques were introduced along with various sensors, bio-markers, and health records to further aid with analysis and visualization towards disease detection. In [7], the authors introduce Syndrome, their proposed framework that addresses hardware security concerns for MIoT devices. This work leverages threat modeling and IDS to tackle malware activity detection and operating system, OS, interrupts through real-time MIoT monitoring. In [8], the authors propose an end-to-end solution for RPM, including data and video monitoring, notification of abnormal conditions, and control of smart appliances. Additional contributions of this work include interoperability of various IoT sensors and ease of access, and a cross-platform user experience. The remainder of this paper is structured as followed. In section II, we discuss research challenges, section III, the main ideas are outlined, section IV, the proposed research plan is introduced, and section V, expected results are presented. II. RESEARCH CHALLENGES Our current research objectives have yielded several questions, otherwise referred to as research challenges; they are as follows: • How do we ensure device interoperability, the security of various communication protocols after configuration (i.e., Bluetooth, WiFi, Zig-Bee)? • Will authorized voice-activated devices pose an additional threat to medical devices if they are interconnected? • How do we defend against unauthorized attempts to access networked devices? • How can we reassure consumers of smart medical technologies, as well as smart home users, that their personal data will not be mishandled? • What additional threats, if any, might be posed to this architecture as well connected MIoT devices and data if non-medical related home automation devices and systems become infected (i.e. home router, refrigerator, door lock)? III. EXPLANATION OF THE MAIN IDEAS This research’s primary motivation is to protect individuals who rely on smart medical devices for daily use and various health concerns and are interested in introducing these devices onto their home network. In table I, the related work presented in section I are compared to the main ideas presented below: • Privacy for Home Users – Unauthorized personnel should not be granted access to or know that the medical devices that exist on the network or the patients who utilize them. • Security of Personal Data – Ensuring unauthorized personnel will not have access to patient sensitive information ranging from medical record numbers to email addresses, physical traits, or genetic make-up. • Prevent Modification of Patient Data – If unauthorized personnel can view patient sensitive information (i.e., test results, medication, etc.), notify authorized users, and perform checks to ensure data integrity. • Cyber Resilience – In the event of a cyber-attack, device functionality should not be affected. • Installing Secure Updates – When the manufacturer of a product performs routine updates, the device(s) should not be rendered vulnerable to attacks or data breaches. • Decommissioning – Once a device is removed from the ecosystem or taken outside of the home, the device, data, and configurations should reset to reduce attackers from accessing or stealing data. IV. PROPOSED RESEARCH PLAN A. Threat Modeling and Intrusion Detection It is evident by the literature searches in table 1, that existing architectures already tackle security and privacy for the MIoT area extensively. However, few architectures may be adaptable for smart home environments as they do not address the main ideas and research questions henceforth presented in this work. Thus, in this phase of our research, we will develop a small-scale testbed, depicted in figure 1, to further evaluate said concepts by implementing and evaluating threat modeling effectiveness. The Raspberry Pi and the WiFi Dongle will serve as a wireless access point or gateway on which the healthcare set-up will rely upon, and devices will be granted internet access and monitored without disrupting the entire home network. The Raspberry Pi will also host the database to collect and store patient-sensitive data for visualization, using an external touch screen. Additionally, an Echo Dot will have access to the database, reporting the most recent readings for home users audibly. In this ecosystem, we will be exploiting two commercial MIoT devices, the Smart Glucose Monitoring System and the Smart No-Touch Forehead Thermometer, as well as the AD8232 ECG Module and a Smart Switch. Additionally, while the concept of Intrusion Detection Systems, IDS, is not entirely novel and several applications currently exist, collecting and analyzing IoMT datasets on networks also that consist of home automation devices is one contribution of this work. Stemming from this network configuration and data capture, we will perform various cyberrelated attacks to assess the impact of interconnected devices and resources. In this set-up, we will evaluate a scenario in Authorized licensed use limited to: Carleton University. Downloaded on June 19,2021 at 21:51:22 UTC from IEEE Xplore. Restrictions apply. TABLE I COMPARISONS OF RELATED WORKS BASED ON THE MAIN IDEAS for consumer safety, as the modification of devices or data may lead to adverse health effects, including end of life for end-users. As evident in ”Security and Privacy Issues with IoT in Healthcare” and as early as 2015, a researcher was able to successfully administer a lethal drug dosage via an infusion pump, further emphasizing the dangers of device misusage and medical malpractice. Beginning with the Smart Glucose Monitoring System and the Thermometer, we will examine several scenarios, two of which focus on the glucose meter and are depicted in figures 2 and 3, with the primary objectives being: • Fig. 1. Small Scale Testbed Network Configuration which a patient is interested in sending their sensitive medical information and device data readings to their HCP. Thus, interference with the network by malicious personnel may look like the following. • • • User(s) credentials being compromised, Home automation and/or medical devices or services being unavailable, modified, or disrupted, Incorrect readings, which may lead to a misdiagnosis or even death The first item from above poses a serious concern for consumers who use the same access credentials across multiple platforms and devices. Once an attacker has the credentials, they may have access to everything, including banking, medical record number, etc. However, the second item has the potential to affect both smart home users and MIoT device holders as the confidentiality, integrity, and availability are all called into question. The final item presents the greatest risk • Attacking the Bluetooth protocol to disrupt the devices’ typical behavior and Exploiting web-based applications associated with smart devices and exposes additional threats by allowing attackers to export data readings, as a PDF, CSV, or Excel spreadsheet, as well as transmitting the data over the internet. Next, utilizing the AD8232 ECG Module, we will develop our own medical IoT device to become better acquainted with the role manufacturers play in the development or lifecycle of a product, from specifying the communication protocol(s) [9], [10] [11] to managing data, and installing updates. Exploring the manufacture’s role offers several benefits, including incorporating security and privacy in the design phase. Additionally, this step will allow us to investigate the behavioral patterns of various and interconnected communication protocols. Lastly, we will evaluate the issues that may arise for devices that require power from an external outlet, non-battery operated. In this case, we will assume the patient will be using a smart switch to supply power to said devices. If a malicious person interferes with this device, connected technologies’ performance may also be rendered vulnerable to attacks. Thus, Authorized licensed use limited to: Carleton University. Downloaded on June 19,2021 at 21:51:22 UTC from IEEE Xplore. Restrictions apply. Fig. 2. Man-in-the-Middle (MitM) Attack Scenario Fig. 3. Social Engineering Attack Scenario the final solution must also be able to detect and defend against attacks aimed towards the smart home’s interconnectivity as well. B. Full-Scale System Framework Bringing IoT into the healthcare sector will raise complications because these organizations are dealing with networks outside of the secure networks they have created. Having a medical device connected to a home network, cellular signals, and public WiFi, and feeding data back to a hospital’s network presents another critical area for weaknesses [4]. Therefore, in this work, we will consider only the smart home user’s impact, not the entire system. In our approach, we consider a fully automated smart home, and we will isolate the smart medical devices on the network using a wireless access point as a gateway, same as in the first phase. We will profile the connected MIoT devices and collect data using a secure database and server, respectively. Here we will also be interested in developing a graphical user interface, GUI, to create user profiles and assess user preferences to incorporate with data management strategies. Our architecture is depicted in figure 4, and in the following sections, we will further discuss the considerations for each of the core concepts. 1) Notification Center - The notification center will be an available feature via the client-side application; if a STRIDE event occurs, the following alarms will be presented: • Cyber Incident Response – This alarm will be triggered if a malicious actor aims to access the network remotely via bypassing the isolated gateway, providing incorrect user login credentials, interacting with the Alexa voice assistant, or attempting to view or manipulate the secure database. When the alarm is sounded, the secure database will encrypt the last trusted instance for data collected before removing itself from the network and critical systems will be taken offline until the malicious person or code is removed from critical systems as well as the network. • Physical Incident Response – This alarm will be triggered when there is a physical event in which a device has been altered, removed, or further tampered with physically. If the device is removed from the network, all previous data will remain on the device server; however the devices themselves will be factory reset in order to prevent data breaches. However, if the devices have instead been tampered with, a report will be generated for the end-user and the last saved data will be saved on the device server; no further data will be considered until the device has returned to normal functionality. 2) Medical IoT Device Server –This server will be responsible for profiling and hosting the connected smart medical devices. Here we address access control, a method for effectively monitoring the access of resources and prevention of unauthorized flow of information [12]. 3) Secure Database – In our approach, we will be addressing the security preferences of home users as well as users of smart medical devices, thus the secure database will be in charge of the following: • Data Storage – data collected at the IoT Device server will be assessed using privacy awareness preferences before being sent the secure database for storage • Data Visualization – the data collected on the secure database will be sent to the touch screen, via the application, to visualize patient vitals • Audible Data – the Echo Dot voice assistant will also have access to the secure database with limitedread-only access; this will allow for recent data to be heard audibly at the designated user’s request 4) Trusted Server – This server will be the centralized server responsible for performing secure handshakes between the device manufacturer, hospital staff, and their respective systems and services. This server will also be responsible for ensuring the information is what is says it is, as it is coming from a trusted source. In this design, we recommend doctors allocate a designated computer(s) to host the software and communicate with patients. Additionally, since human error is also a great concern Authorized licensed use limited to: Carleton University. Downloaded on June 19,2021 at 21:51:22 UTC from IEEE Xplore. Restrictions apply. Fig. 4. Proposed Large Scale Network Configuration for IoT devices, doctors will only be allowed limited access to view patient sensitive data, monitor vitals, as well as provide medication recommendations; not supply medication dosages or operate machinery. On the user’s end, once the doctor provides recommendations, they may review and either accept or decline recommended medications and dosages as well as other treatments before being disseminated to users. The medical physician will then receive a report of recommendations that were accepted or ignored by users. Manufactures of IoT devices will also be granted limited access to the network configuration to provide updates securely and only their specific networked device. The trusted server will be in charge of verifying and validating the updates provided by manufacturers, alerting users of the bugs being fixed and the critical systems that may be affected during the update. V. EXPECTED RESULTS Ultimately, manufacturers of smart medical devices should adopt a security-by-design and privacy-by-design method for the development of their devices [4]. However, many of the devices on the shelves today do not adhere to this standard; thus, what is to be said for consumers who have already purchased smart medical devices, especially users who intend to utilize their devices in smart home ecosystems? At the completion of this research, we expect to provide a secure system architecture independent of whether or not a device has been designed with security and privacy in mind, which will reassure consumers of smart medical devices as well as smart home users. The completed solution will address privacy awareness concerns for MIoT devices and provide securelimited access for manufacturers to supply secure updates and medical professionals to monitor patient vitals and provide health recommendations remotely for smart home users. ACKNOWLEDGMENT This work was supported by the Center for Reverse Engineering and Assured Microelectronics (CREAM) research lab at Morgan State University. We would also like to thank Myles Wright-Walker as well as the reviewers for their additional insight. R EFERENCES [1] A. J. Brush, J. Albrecht, and R. Miller, “Smart homes,” IEEE Pervasive Computing, vol. 19, no. 2, pp. 69–73, 2020, doi:10.1109/MPRV.2020.2977739, accessed = 2019-08-31. [2] I. Chiuchisan, H. Costin, and O. Geman, “Adopting the internet of things technologies in health care systems,” in 2014 International Conference and Exposition on Electrical and Power Engineering (EPE), 2014, pp. 532–535, doi:10.1109/ICEPE.2014.6969965, accessed = 2019-08-31. [3] A. Chacko and T. Hayajneh, “Security and privacy issues with iot in healthcare,” EAI Endorsed Transactions on Pervasive Health and Technology, vol. 4, no. 14, 2018, accessed = 2019-08-31. [4] A. Seeam, O. S. Ogbeh, S. Guness, and X. Bellekens, “Threat modeling and security issues for the internet of things,” in 2019 Conference on Next Generation Computing Applications (NextComp). IEEE, 2019, pp. 1–8, doi:10.110/NEXTCOMP.2019.8883642 accessed = 2020-07-20. [5] I. Psychoula, L. Chen, and O. Amft, “Privacy risk awareness in wearables and the internet of things,” IEEE Pervasive Computing, vol. 19, no. 3, pp. 60–66, 2020, doi:10.1109/MPRV.2020.2997616 accessed = 2020-08-20. [6] M. Hassanalieragh, A. Page, T. Soyata, G. Sharma, M. Aktas, G. Mateos, B. Kantarci, and S. Andreescu, “Health monitoring and management using internet-of-things (iot) sensing with cloud-based processing: Opportunities and challenges,” in 2015 IEEE International Conference on Services Computing, 2015, pp. 285–292, accessed = 2019-08-28. [7] N. Sehatbakhsh, M. Alam, A. Nazari, A. Zajic, and M. Prvulovic, “Syndrome: Spectral analysis for anomaly detection on medical iot and embedded devices,” in 2018 IEEE International Symposium on Hardware Oriented Security and Trust (HOST), 2018, pp. 1–8, accessed = 2019-08-30. [8] H. Moustafa, E. M. Schooler, G. Shen, and S. Kamath, “Remote monitoring and medical devices control in ehealth,” in 2016 IEEE 12th International Conference on Wireless and Mobile Computing, Networking and Communications (WiMob), 2016, pp. 1–8, accessed = 2019-08-31. Authorized licensed use limited to: Carleton University. Downloaded on June 19,2021 at 21:51:22 UTC from IEEE Xplore. Restrictions apply. [9] S. Shebi Ahammed and B. C. Pillai, “Design of wi-fi based mobile electrocardiogram monitoring system on concerto platform,” Procedia Engineering, vol. 64, pp. 65 – 73, 2013, international Conference on Design and Manufacturing (IConDM2013), accessed = 2020-0902. [Online]. Available: http://www.sciencedirect.com/science/article/ pii/S1877705813015919 [10] A. AKBULUT, “Implementation of zigbee (ieee 802.15. 4) based wireless ecg measurement system,” Uluslararası Mühendislik Araştırma ve Geliştirme Dergisi, vol. 9, no. 3, pp. 43–53, 2017, doi:10.29137/umagd.360670 accessed = 2020-09-02. [11] H. Yang and J. Chai, “The study and design of a wireless ecg monitoring system,” Biomedical Instrumentation & Technology, vol. 46, no. 5, pp. 395–399, 2012, doi:10.2345/0899-8205-46.5.395 accessed = 2020-0902. [12] J. Qiu, Z. Tian, C. Du, Q. Zuo, S. Su, and B. Fang, “A survey on access control in the age of internet of things,” IEEE Internet of Things Journal, vol. 7, no. 6, pp. 4682–4696, 2020, doi10.1109/JIOT.2020.2969326 accessed = 2020-09-02. Authorized licensed use limited to: Carleton University. Downloaded on June 19,2021 at 21:51:22 UTC from IEEE Xplore. Restrictions apply.