This proposal seeks funding for the Center for Autonomic Computing (CAC) studies conducted by the University of Florida site (lead) and the University of Arizona site. Funding Requests for Fundamental Research are authorized by an NSF approved solicitation, NSF 10-507. The solicitation invites I/UCRCs to submit proposals for support of industry-defined fundamental research.
This project proposal focuses on autonomic computing and systems, applicable to the important areas of security, fault management, and data centers. Planned are the prototypes to be deployed on test bed environments driven by the requirements from industry. The most interesting part of the proposal are applications to SCADA environments, which integrate monitoring, multi-level behavior analysis, decision fusion and risk analysis relative to security of SCADA environments. Other areas include autonomic peer-to-peer systems, as well as study related to the data centers, both at the application layer and data center hardware, where focus is on robustness.
The successful completion of this project will represent a significant step toward the design and deployment of highly secure SCADA systems and networks. In addition to security, the autonomic detection of anomalies in applications or subsystems of a datacenter addresses inefficiencies in data center design by eliminating the dependence on over-provisioning, which is a resource-inefficient strategy for ensuring quality of service by trying to offset any sub-system failures and malfunctions. The project has also potential for broader impact on data-intensive applications that leverage the availability of ad-hoc desktop grids for high-throughput computing.
This project has made contributions in autonomic approaches for computing systems that are able to self-protect against intrusion, to self-organize in overlay networks to achieve high-throughput data transfers, and to detect and classify thermal anomalies within data centers. In one thrust of the project, a testbed to study and simulate the various available techniques for securing and protecting Supervisory Control and Data Acquisition (SCADA) systems against a wide range of cyber-attacks has been developed. The critical infrastructures of our society are in the process of being modernized. Most significantly impacted are the industrial control systems through replacement of old electromechanical systems with advanced computing and communication technologies. This modernization has introduced new vulnerabilities to those infrastructures. Securing critical infrastructures is a challenging research problem, as these control systems were not designed with security in mind. SCADA systems consist of human-machine interfaces (HMI), historical database management systems (HDBMS) and sophisticated monitoring and control applications. The SCADA system manages a collection of distributed industrial control components including remote terminal units (RTUs) for field sensor and actuator connectivity, programmable logic controllers (PLCs) that perform simple logic processes, and a wide variety of intelligent electronic devices (IEDs) for process data collection and control. Unexpected changes in local heat-generation and -extraction rates due to cooling equipment failures, misconfigurations, and attacks may over time cause large heat imbalances, and unexpected thermal hotspots. Thermal hotspots may also result in a thermal fugue, which is characterized by a continuous increase in the rate of temperature rise. Thermal anomalies, such as unexpected hotspots and fugues, lead to system operation in unsafe temperature regions, increase the server failure rate and the Total Cost of Ownership (TCO) of datacenters. In another thrust of the project, an online autonomic thermal-anomaly detection method that leverages the novel notion of the thermal signature of a datacenter has been developed. The heat-imbalance model is used to estimate approximate intensities and distribution of expected hotspots for a specific workload distribution (the datacenter's unique thermal signature). Autonomic middleware services researched in this project also address data access needs that are increasingly important in enterprise applications that are outsourced for execution in infrastructure-as-a-service, cloud computing platforms. In particular, the project focused on on-demand, piece-wise, cooperative transfer of large, read-only or read-mostly datasets, such as virtual machine images for large-scale cloud provisioning. Effective handling of data transfer and storage can lead to cost reductions in the use of provisioned cloud infrastructures, which charge customers on the basis of bytes stored and transferred over time. In this project, a novel architecture for a self-configuring, autonomic file system service has been researched, and a prototype based on a peer-to-peer (P2P) overlay, BitTorrent data transfer and cooperative caching, and file system in user space (FUSE) bindings has been implemented. Experiments have shown the ability to achieve higher data transfer throughput compared to traditional client/server file systems.
