Mobile Assisted Security Framework for Wireless Sensor Networks

Advances in technology have enabled low-power, low-cost, and small devices called sensor nodes that can compute, sense, and communicate with each other wirelessly. Wireless sensor networks (WSNs) involve a large number of autonomous sensors distributed over an area, and have many applications including military, environment and habitat monitoring, healthcare, home automation, and traffic control. Sensors are unattended once they are deployed, and are vulnerable to attacks by adversaries. Therefore, security plays an important role to make these applications practical. However, securing a WSN is challenging because of the unreliable and infrastructureless wireless communication; and the limited resources on the sensor nodes such as low memory, low computational power, and limited energy supplies. Moreover, developing a scalable security infrastructure to handle the large number of sensor nodes is challenging.

In this dissertation, mobile elements (MEs) are used to ease the challenges in WSN security and to mitigate the overhead on the sensor nodes. The main building block of our approach is an ME with location information. Location information is vital for secure protocol development since many protocols make use of location information. Although, Global Positioning System (GPS) can be used outdoors, options for indoors are limited. Therefore, we propose an indoor location management system using infrared (IR) leds and an IR camera. Then, we introduce a mobile assisted key distribution scheme where the ME handles all the overload of key distribution requiring minimal resources at the sensor nodes for key management. Thirdly, we show how to coordinate a team of MEs to guarantee continuous connectivity among them since connected MEs help develop security protocols. Next, a secure data collection scheme using transmit-only sensors and an ME is proposed. Transmit-only sensors eliminate most of the vulnerabilities including attacks against routing protocols since a transmit-only sensor does not have a receiver circuitry. We demonstrate that MEs and transmit-only sensors can be used for data collection in hostile environments. Since the MEs maintain essential information, they are attractive to adversaries. To address this issue, we present a method to verify the traversal of the ME by node encounters. Furthermore, we develop a data collection method resilient to node failures or packet dropping attacks using trajectory routing. Finally, we focus on the detection of sinkhole and wormhole attacks using MEs in networks which employ static routing. Proposed method uses counters for data flow on the sensors. Proposed protocols are evaluated by extensive simulations and shown to be practical.