Ultra-dependable, High-performance, Real-time Signal Processing

The Active Nodal Task Seeking (ANTS) approach to the design of multicom-puter systems is named for its basic component: an Active Nodal Task-Seeker (ANT). In this system, there is no load balancing or load sharing, instead, each ANT computing node is actively nding out how it can contribute to the execution of the needed tasks. A run-time partition is established such that some of the ANT computing nodes are under exhaustive diagnosis at any given time. An ANTS multicomputer system can achieve a mean time to failure of more than 20 years with just 8 computing nodes and 3 buses, while the minimum requirements are 3 computing nodes and 1 bus, and with a worst case computing node failure rate of 5 10 ?4 per hour. In this nal report, we present the many ndings during the project duration. First, the rst version of ANTS kernel has been implemented on a network of UNIX workstations. Employing TCP/IP protocol for communication via Ether-net LAN. The second version of ANTS kernel will exist in a workstation in the form of a UNIX daemon. In such a case, there will be no performance shortfall observable by the casual workstation user when the kernel is not engaged in any ANTS job. Based on these implementations, we have successfully demonstrated the proposed fault tolerant techniques via fault injection experiments. The system detects faults and recovery itself as expected within seconds. The dependability goal predicted in the theoretical model is thus validated. We then program and execute several well-known algorithms in ANTS distributed computing environment to demonstrate the high-performance features. Algorithm designs for ANTS environment are also discussed. This work has been motivated by the need to develop high-performance mul-ticomputer systems for radar, active and passive sonar, and electronic warfare that can provide ultra-dependable performance for more than 20 years without eld repairs. We argue that high performance is also an attribute of an ANTS computing system, because the overhead of dynamic task scheduling is reduced and because eecient use is made of the available processing resources. We have successfully demonstrated that both dependability and high-performance goals of our proposal have been achieved. This forms a basis for a practical real-time systems.