An agent framework for survivable network systems

Abstract

Large network systems elevate the levels of e f iciency and effectiveness of organisations by improving systems integration. This improved integration of computer networks and information systems is accompanied by increased risks of intrusion and compromise. Survivability of networks and information systems in face of these risks becomes an importantaspect in the effective functioning of a n organisation. The unbounded nature of today's networks and the Internet make it impossible to fore see all possible risks and cover the systems against all possible attacks. In this scenario, it becomes necessary that the network be designed to resist possible attacks, recognise attacks and recover from attacks. The open-ended nature of this problem requires that flexible architectures be employed t o improve survivability. Agent technology is a highly flexible and robust paradigm f o r building large scale distributed systems. In this paper, we present a n agent-based framework f o r Survivable Network Systems. 0-7803-5258-0/99 $10.00 0 1999 IEEE

Intelligent Control for Future Autonomous Distributed Sensor Systems

“…Survivability addresses the ability of a system to respond to attacks, failures, and accidents [15]. Survivability depends on four key properties: a) resistance to threats, b) recognition of threats and the extent of any damage suffered, c) recovery of full and essential services after damage, and d) adaptation and evolution to reduce effectiveness of future threats.…”

“…Multi-agent systems provide a level of abstraction that enhances computational efficiency, reliability, extensibility, maintainability, responsiveness, flexibility, and reuse. With these properties, multi-agent system technology appears to be an ideal candidate to realize autonomous distributed system designs and provides a framework that may be useful in achieving survivability [15], [45].…”

“…For repeated trials, using (14) the gateway level detection and false alarm probabilities are given by PDg=r Pd (1-P)fY-k (15) PFg k=r, (fk…”

“…If the nfg decisions that were reported to the fusion center were used to carry out a gateway level detection using nfg repeated trials instead of being used to make a decision at the field level, the probability of detection PDg* for the gateway could be computed using (15). However, the nfg decisions reported to the fusion center were not used to carry out a gateway level detection.…”

“…For Pj = 0.216970 and Pd = 0.821320, the probabilities (PDg and PFg) of getting exactly a) k detections and false alarms out of n trials and b) k or more detections and false alarms out of n trials were computed using (15) and (16) for different numbers of trials. These probabilities were examined to allow the threshold, Tg, and number of repeated trials, n, to be chosen to meet the specified PDg > 0.9 and PFg < 0.025 (Table 1).…”