<p>Despite the many security and privacy concerns associated with Internet-of-Things (IoT) devices, we continue to be barraged by new IoT devices every day. These devices have infiltrated almost every aspect of our lives, from government and corporations to our homes, and now, on and within our person, in the form of smartphones and wearables. These personal IoT devices can collect some of the most intimate pieces of data about their user. For instance, a smartwatch can record its wearer's heart rate, skin temperature, physical activity, and even GPS location data. At the same time, a smartphone has access to almost every piece of information related to its user, including text messages, social media activity, web browser history, and application-specific data. Due to the quantity and quality of data these personal IoT devices record, these devices have become critical sources of evidence during forensic investigations. However, there are instances in which digital forensic investigators need to make doubly sure that the data obtained from these smart devices, in fact, belong to the alleged owner of the smart device and not someone else. To that end, this dissertation provides the first look at using personal IoT device handling as a user identification technique with machine learning models to aid forensic investigations. The results indicated that this technique is capable of significantly differentiating device owners with performance metrics of .9621, .9618, and .9753, for accuracy, F1, and AUC, respectively, when using a smartwatch with statistical time-domain features. When considering the smartphone performance, the performance was only marginally acceptable with accuracy, F1, and AUC values of .8577, .8560, and .8891, respectively. The results also indicate that female users handled their devices notably differently from male users. This study thus lays the foundation for performing user identification during a forensic investigation to determine whether the smart device owner did, in fact, use the device at the time of the incident.</p>
| Identifer | oai:union.ndltd.org:purdue.edu/oai:figshare.com:article/23739981 |
| Date | 07 August 2023 |
| Creators | Shinelle Hutchinson (16642559) |
| Source Sets | Purdue University |
| Detected Language | English |
| Type | Text, Thesis |
| Rights | CC BY 4.0 |
| Relation | https://figshare.com/articles/thesis/Leveraging_Personal_Internet-of-Things_Technology_To_Facilitate_User_Identification_in_Digital_Forensics_Investigations/23739981 |
Page generated in 0.0022 seconds