- Professor Ravishankar K. Iyer
- Zbigniew Kalbarczyk (Research Scientist)
- Keith Whisnant (Graduate Student)
- Jun Xu (Graduate Student)
- Nick Breems (Graduate Student)
The goal of our research is to investigate fault-tolerant and secure communication in a wireless/wireline environment, e.g., an ad-hoc sensor network. Distinguishing characteristics of an environment such as wireless sensor network encompass low bandwidth, relatively low reliability and security, and variable delay in data transmission. As a result, disconnection is a common phenomenon. To ensure robustness of ad-hoc networks, we need mechanisms (e.g., reliable mobile agents) to detect and to repair (recover) network partitions/holes. System must be able to determine when areas of the network have failed and security requires some way to validate that packets came from a legitimate host.
In our approach we: (1) explore and prototype a transparent, high-availability framework for supporting client-server applications operating over wireline and/or wireless networks, (2) investigate issues and solutions (e.g., protocols) in supporting reliable and secure communications in wireless (e.g., sensor networks), and (3) develop Remote Vehicle testbed, to investigate and test real-time, secure and fault-tolerant control. The specific research areas include:
- Providing software framework for robust connectivity to client-server applications, and protecting the applications and the service against various failure modes by supporting transparent detection and recovery from errors.
- Characterizing capabilities and limitations of fault-tolerant, real-time mechanisms for remote control.
- Investigating mechanisms and/or protocols for error/intrusion detection and recovery in sensor networks.
Recent Accomplishments and Activities
We have proposed and prototyped a transparent, software framework for providing robust connectivity to client-server applications and protecting applications and services against various failure modes. The recovery procedure includes re-establishing the lost connections between the client and the server and recovering the application processes. Our scheme emphasizes availability, while preserving mobility and persistent connectivity over wireless networks. One of the distinguishing features of the framework is the transparency with which it offers the services. Both ends of the connection (server and client) are unchanged (the solution is implemented in the user-level code).
The proposed framework builds upon the ARMOR processes (Adaptive Reconfigurable Mobile Objects for Reliability). The scheme is proxy-based and introduces two redirection proxies between the client and server processes. These proxies are a special service provided by the ARMORs to help maintain track of the open TCP connections between the clients and servers. A failure at either end of the connection is detected by ARMORs. The recovery phase uses the information stored in the redirection proxy elements to restore the connections that existed before the failure. For an application demanding persistent connectivity in the wireless environment where connections might drop (due to faulty units, poor placement or enemy activity), this scheme keeps the connections alive during periods of unstable connectivity of the client.
We have developed prototype of an ARMOR-based Remote Vehicle testbed, to investigate real-time, fault-tolerant control. The testbed provides wireless network simulator as an in-line, bridge-level simulator that can customize network performance to match any specified target. Using the simulator one can mimic variation in bandwidth, delay in data transmission, background noise and different failure modes of application and/or network. We have started to take initial measurement using the testbed.
We have started (in collaboration with the University of Virginia) to explore issues and solutions for providing secure communications in a wireless sensor environment. In particular we focus on:
- Intrusion and error/failure detection in a sensor network – detecting sensor failures, distinguishing between a failure and an attack, recovering from failures/attacks, and understanding impact of transient errors on system/network security.
- Developing algorithms for a region mapping – selecting criteria for detecting jamming, developing mapping algorithms.
- Providing protocols and algorithms for authenticated broadcast – methods to validate origin of a message, use the natural redundancy of broadcasting to support information authentication.
Real-time fault tolerance:
- Apply the ARMOR-based software framework for providing real-time, fault-tolerant control of the remote vehicle.
- provide fault tolerance (detection and recovery) services to the control software on both the server and the on-board client
- protect the wireless connection, and provide reliable near-real-time transfer of data over an encapsulated tunnel.
- protect wireless video channel between the vehicle and the control server.
- Conduct measurement-based analysis of controllability of the remote vehicle in presence of application and/or network errors/failures and variability in the bandwidth of the wireless network.
- Explore mechanisms/algorithms for identifying jammed or totally failed area in the wireless network,
- Avoid routing through jammed or unresponsive areas,
- Propose methods for repairing/recovering failed areas, and
- Provide inexpensive authenticated broadcast channel between base station and sensors (challenges – solutions using cryptography not feasible in sensor network due to limited network bandwidth, battery power and computation/memory capabilities).
For additional information on the MURI project, click here.
- S. Chen, J. Xu, R.K. Iyer, K. Whisnant, “Evaluating the Security Threat of Firewall Data Corruption Caused by Instruction Transient Errors,” in Proc. of Conference on Dependable Systems and Networks, DSN’02, June 2002, pp.495-504.
- K. Whisnant, R.K. Iyer, P. Jones, R. Some, D. Rennels, “An Experimental Evaluation of the REE SIFT Environment for Spaceborne Applications,” in Proc. of Conference on Dependable Systems and Networks, DSN’02, June 2002, pp.585-594.