applications. For example, the RSA algorithm is homomorphic with respect to multiplication, but it is not homomorphic with respect to addition. Moreover, PHE schemes have massive commu-nication overheads because data owners and the service provider are required to exchange data several times to perform computations that can-not be performed in the encrypted domain, or for verification purposes. In contrast, FHE schemes require the encrypted data to be sent only once to the server provider, (the Cloud Service Pro-vider (CSP) in our case), and all computations are performed securely in the encrypted domain without the need to interact with data owners during the analysis tasks. However, due to the large computational overhead, most FHE schemes are still far from being ready to use in practical applications. While recent developments in FHE schemes have sped up performance, only with the development of the homomorphic encryption library (HElib) using the BGV scheme  has it been possible to build secure applications based on FHE. We take advantage of FHE to build an independent privacy-preserving healthcare mon-itoring and prediction model that can perform Security mechanisms, such as authentication, authorization, policy integration, access con-trol and cryptography, must play a critical role to preserve the privacy of PHI. Storing massive volumes of sensitive medical data in TTP cloud-based storage is susceptible to leakage or loss.
IEEE Communications Magazine • January 2019125encrypted-based analysis tasks and that eliminates most of the security and privacy vulnerabilities of other cryptographic techniques.PRIVACY-PRESERVINGREAL-TIMESURVEILLANCEANDSMARTDECISION-MAKINGSYSTEMIn this section, we describe the architecture of our system and illustrate how different entities interact in a privacy-preserving manner. We first provide an overview of the system architecture and then describe the security mechanism that is used to protect the privacy of the patient data during different processing stages. Next, we explain how we deploy the change detection and abnormality prediction model. Since the compu-tation of homomorphically encrypted data is very CPU intensive, we also show how multiple virtual machines can be utilized to parallelize the process and to improve the execution time significantly.SYSTEMARCHITECTUREThe proposed system provides a privacy-preserv-ing cloud-based real-time change detection and abnormality prediction framework for multiple vital signs of a patient. The system architecture has three main entities, as follows:• Smart Community Resident (SCR): Multiple vital signs data of community residents can be aggregated and sent directly to the cloud-based storage.• Cloud Storage (CS): Used to store SCR’s data in an encrypted form.