With the rapid growth in the number of IoT devices being added to the network, a major concern that arises is the security of these systems. As these devices are resource constrained, safety measures are difficult to implement on the edge. We propose a novel approach for the detection of IoT device attacks based on the use of formal modeling and mutation testing. Namely, we model the behavior of small IoT devices such as motion sensors and RFID reader as state machines with timeouts. We also model basic IoT attacks; namely, battery draining, sleep deprivation, data falsification, replay, and man in the middle attacks, as special mutants of these specifications. We also consider tests for detecting actual physical device manipulation. Mutation testing is then used to derive tests that distinguish these attacks from the original specifications. The behavior of these mutants is tested in real environment by running the tests on them. Our experiments show that derived the number of attack mutants and tests is small and thus these tests can be executed many times with limited overhead on the physical device. Consequently, our approach is not deterred by related high costs of traditional mutation testing. In addition, we also show that tests derived by our method which cover all IoT attacks do not provide good coverage of mutants derived using traditional mutation code-based operators and this indicates the need of using our method. A framework that implements our approach is presented along with some other relevant case studies.
View on arXiv