This project focuses on an important kind of networked embedded systems called sensor networks. Such networks have become popular platforms for continuous sensing and analysis of physical environments, leading to a better understanding of natural phenomena, civil infrastructures, animal habitats, and other important scientific and engineering issues. The sensor data can also be used to improve environmental protection, infrastructure safety and energy efficiency, to name a few of the potential applications.

The ad-hoc and dynamic nature of networked embedded systems make their communication protocols complex. This research uses the methodology of continuous error monitoring for continued improvement of reliability after the deployment of sensor networks. The project builds a framework consisting of compiler-based tools and software techniques for the detection, diagnosis and correction of programming errors on sensor networks. The nature of the sensor-network applications requires the hardware components and software protocols to be small and resource-constrained. The project therefore makes resource efficiency one of the critical design criteria.

The success of this project should contribute substantially to the reliability of sensor networks whose potential applications are very broad. The project also engages a broad community of students in various disciplines, through Honor Seminars, special mentoring programs and undergraduate research opportunities such as Purdue’s EPICS program (Engineering Projects in Community Services).

The project provides resilience to wireless networks of mobile ad hoc and sensor (MAHAS) nodes, which are vulnerable to a wide range of security attacks. These attacks could involve eavesdropping, message tampering, or identity spoofing, that have been addressed by customized cryptographic primitives for encryption and authentication. Alternately, the attacks may be targeted to the control or the data traffic in wireless networks. Since the networks are resource constrained (bandwidth, energy, or processing), providing detection and countermeasures to such attacks often turn out to be more challenging than in wired networks. The project is developing protocols for detecting, diagnosing, and mitigating one class of attacks, namely, those that affect the control traffic. Typical examples of control traffic are routing, monitoring the liveness of nodes, and system management. It is critical to guarantee the fidelity of control traffic since disrupting it can hamper the data traffic. For example, if a malicious node manages to interpose itself in an established route between two legitimate nodes, it can disrupt the data traffic by selectively dropping the data packets. Such attacks are often difficult to detect and can be launched without the need for cryptographic keys. The research presents a technique called local monitoring whereby nodes oversee part of the traffic going in and out of its neighbor nodes. The project makes neighbor monitoring feasible in resource constrained environments and mitigates the effect of the malicious node through isolation, either locally or globally using a distributed protocol. The work will enable the deployment of MAHAS networks for critical secure applications. We expect application of the results to two testbeds for environment monitoring (water quality and pharmaceutical manufacturing) that we are currently building.

Wireless mesh networks are emerging as a promising robust low-cost network architecture able to provide increased coverage and larger bandwidth, resulting in higher quality of service and information availability. Many distributed applications provided on wireless mesh networks enable collaborations and sharing of information. Such applications benefit from scalable, robust and secure group services such as one-to-many or many-to-many multicast and distributed data access. Group services, in turn, require support from infrastructure protocols such as routing, or security mechanisms such as authentication, access control and key management.

The goal of this project is to advance state-of-the-art group-oriented services in the context of wireless mesh networks. The project presents new formulations of distributed problems capturing the particularities of wireless mesh networks and the interactions between security, availability and scalability. It also studies the viability and limitations of cross-layer design as a new paradigm of building secure network services. Innovative results consist of scalable one-to-many and many-to-many wireless multicast protocols that provide support for efficient group communication, scalable data sharing algorithms that are robust to malicious behavior, high-throughput, robust and secure routing that assist applications to achieve high-performance, scalable key management protocols and authentication mechanisms enabling decentralized infrastructure access services. This project contributes to the education of the next generation of secure systems designers, generating interactions between the distributed systems, security and wireless networks research areas.