Security- and Privacy-Aware IoT Application Placement and User Assignment


Applications for the Internet of Things (IoT) may use, beyond the IoT devices themselves, also edge and cloud resources. Thus, the modules of an application can be placed on a variety of nodes with different capabilities in terms of security, trustworthiness, and capacity. Application modules may exist in multiple instances. This makes it possible to assign users to the most appropriate module instances, taking into account requirements on security, privacy, and latency. There is a non-trivial interplay between application placement decisions and user assignment decisions. For example, if a certain user is assigned to a module, then that module may not be allowed to be placed on nodes not trusted by the user. However, most existing research neglects this interplay and its implications on security and privacy. In this paper, we address the joint problem of application placement and user assignment. Beside capacity and latency constraints, we consider several types of security and privacy constraints: (i) module-level location constraints, (ii) user-level location constraints, (iii) co-location constraints, and (iv) k-anonymity constraints. We formalize the problem and develop an algorithm to solve it using quadratically constrained mixed integer programming. We demonstrate the applicability of the proposed approach by applying it to an IoT system in the smart home domain. Controlled experiments on problem instances of increasing size show that the algorithm can solve even large problem instances in acceptable time.

Computer Security. ESORICS 2021 International Workshops
Zoltan Mann
Zoltan Mann
Associate Professor