Place: BJIM Level 5, School of Computer Sciences
The proliferation of smartphone users has introduced a sensing technology called Mobile Crowd Sensing. This emerging sensing trend leverages smartphones with sensing and communication capabilities to acquire real-time and continuous data from humans and the physical environment. Unlike traditional Wireless Sensor Networks, MCS exploits the mobility attribute in smartphones to offer large-scale sensing at low cost. However, challenges such as battery limitation in smartphones, insecure data fusion, data prioritization, and privacy of users hinder the full adoption of this emerging technology. As regards security, MCS applications are vulnerable to different attacks. To validate this claim, this research work conducted vulnerability analysis on 35 Android-based sensing applications using Burp Suite (a penetration testing tool). Results from the preliminary study showed that MCS applications lack secure fusion and data priority techniques. These lapses lead to transmission of sensitive data in plaintext by these applications without appropriate encryption and authentication. This result signifies a 100% probability that users’ privacy can be violated via these applications. Also, data modification was possible in 33 out of the 35 sensing applications tested. With this result, there is a 94% probability of unreliable data transmission by these applications. In an effort to ensure secure data fusion in mobile crowd sensing applications, this research work proposes a method to accurately classify sensor data using Support Vector Machine Algorithm (SVM) algorithm as well prioritize data using Analytic Hierarchy Process (AHP). Data with high priority (which are location-based data from GPS sensor) are encrypted and authenticated using Advanced Encryption Standard 256-Galois Counter Mode (AES256-GCM) algorithm while only authentication is performed on motion-based data with low priority.