Evaluating APS Ecosystem Security : Novel IoT Enabled Medical Platform for Diabetes Patients

Abstract: Computing technology has been getting more reliable and cheaper every year for the past several decades. Consequently, IoT devices have now become a part of medical technology. One example of this is a new open-source technology that has emerged for type-1 diabetes patients, which regulates the patients’ blood glucose levels. One component of this open-source system is AndroidAPS, a mobile application that manages and controls the system by communicating with the two other components: a blood glucose sensor and an insulin pump. Another component is Nightscout, a web application for monitoring T1D patients. Together they form the APS ecosystem that automatically i) reads blood glucose values, ii) syncs the data with Nightscout, iii) stores patient information into Nightscout database, iv) calculates optimal treatment plan, and v) regulates pump for the ideal insulin intake. The whole system bears several critical assets to guarantee patient health. In this thesis, the security of a representative APS ecosystem is studied. We found 5 critical vulnerabilities in the ecosystem: 1) an XSS vulnerability in the web application due to ineffective input sanitization which lead to stealing administrator password from browser cache, 2) highly sensitive patient data is open to public by default, 3) the web application login mechanism, where all system data is managed, is weak against password guessing attacks together with 4) insecure GET requests used for authentication, and lastly 5) any type of database hijack does not trigger any alarms for Nightscout. Successful attacks result in malicious values synchronized from Nightscout to tamper correct insulin delivery calculations. Possible attack scenarios, devised from existing vulnerabilities in this work, show how an attacker can physically harm their victims through their internet-connected insulin pump.