The Internet of Things (IoT) is getting popular everyday around the world. Given the endless opportunities it promises to provide, IoT is adopted by various organizations belonging to diverse domains. However, IoT’s “access by anybody from anywhere” concept makes it prone to numerous security challenges. Although data security is studied at various levels of IoT architecture, breach of data security due to internal parties has not received as much attention as that caused by external parties. When an organization with people spread across multiple levels of hierarchies with multiple roles adopts IoT, it is not fair to provide uniform access of the data to everyone. Past research has extensively investigated various Access Control techniques like Role Based Access Control (RBAC), Identity Based Access Control (IBAC), Attribute Based Access Control (ABAC) and other variations to address the above issue. While ABAC meets the needs of the growing amount of subjects and objects in an IoT environment, when implemented as an encryption algorithm (ABE) it does not cater to the IoT RDBMS applications. Also, given the query processing over huge encrypted data-set on the Cloud and the distance between the Cloud and the end-user, latency issues are highly prevalent in IoT applications. Various Client side caching and Server side caching techniques have been proposed to meet the latency issues in a Client-Server environment. Client side caching is more appropriate for an IoT environment given the dynamic connections and the large volume of requests to the Cloud per unit time. However, an IoT Cloud has mixed critical data to every user and conventional Client side caching techniques do not exploit this property of IoT data.

In this work, we develop (i) an Attribute Based Access Control (ABAC) mechanism for the IoT data on the Cloud in order to provide a fine-grained access control in an organization and (ii) an Attribute Based Cache Consistency (ABCC) technique that tailors Cache Invalidation according to the users’ attributes to cater to the latency as well as criticality needs of different users. We implement and study these models on a Healthcare application comprising of a million Electronic Health Record (EHR) Cloud and a variety of end-users within a hospital trying to access various fields of the EHR from their Smart devices (such as Android phones). ABAC is evaluated with and without ABCC and we shall observe that ABAC with ABCC provides a lower average latency but a higher staleness percentage than the one without ABCC. However, the staleness percentage is negligible since we can see that much of the data that contributes to the staleness percentage are the non-critical data, thus making ABAC with ABCC an efficient approach for IoT based Cloud applications.