Addressing Insider Threats from Smart Devices

Abstract

Smart devices have unique security challenges and are becoming increasingly common. They have been used in the past to launch cyber attacks such as the Mirai attack. This work is focused on solving the threats posed to and by smart devices inside a network. The size of the problem is quantified; the initial compromise is prevented where possible, and compromised devices are identified.

To gain insight into the size of the problem, campus Domain Name System (DNS) measurements were taken that allow for wireless traffic to be separated from wired traffic. Two-thirds of the DNS traffic measured came from wireless hosts, implying that mobile devices are playing a bigger role in networks. Also, port scans and service discovery protocols were used to identify Internet of Things (IoT) devices on the campus network and follow-up work was done to assess the state of the IoT devices.

Motivated by these findings, three solutions were developed. To handle the scenario when compromised mobile devices are connected to the network, a new strategy for steppingstone detection was developed with both an application layer and a transport layer solution. The proposed solution is effective even when the mobile device cellular connection is used. Also, malicious or vulnerable applications make it through the mobile app store vetting process. A user space tool was developed that identifies apps contacting malicious domains in real time and collects data for research purposes. Malicious app behavior can then be identified on the user’s device, catching malicious apps that were overlooked by software vetting. Last, the variety of IoT device types and manufacturers makes the job of keeping them secure difficult. A generic framework was developed to lighten the management burden of securing IoT devices, serve as a middle box to secure legacy devices, and also use DNS queries as a way to identify misbehaving devices.