Peter N. Marinos

Fuzzy Logic and its Application to Switching Systems

Fuzzy logic deals with propositions which may be ascribed values between falsehood and truth (0 and 1) subjectively in either a continuous or a discrete fashion. This is in contrast to ordinary logic (two-valued or k-valued logic) in which a given proposition is ascribed values objectively using either deterministic or probabilistic approaches.

Cache-aided rollback error recovery (CARER) algorithm for shared-memory multiprocessor systems

Three cache-aided error-recovery algorithms for use in shared-memory multiprocessor systems are presented. They rely on hardware and specially designed cache memory for all their soft error management operations and can be easily incorporated into existing cache-coherence protocols. An example illustrating their use in a multiprocessor system employing Dragon as its cache-coherence protocol is given, and the results of a tradeoff analysis are presented.<<ETX>>

Unification of finite failure non-homogeneous Poisson process models through test coverage

A number of analytical software reliability models have been proposed for estimating the reliability growth of a software product. We present an Enhanced Non-Homogeneous Poisson Process (ENHPP) model and show that previously reported Non-Homogeneous Poisson Process (NHPP) based models, with bounded mean valve functions, are special cases of the ENHPP model. The ENHPP model differs from previous models in that it incorporates explicitly the time varying test coverage function in its analytical formulation, and provides for defective fault detection and test coverage during the testing and operational phases. The ENHPP model is validated using several available failure data sets.

Synchronization of fault-tolerant clocks in the presence of malicious failures

The problem of achieving global clock synchronization in fault-tolerant clocks by preventing so-called multiple cliques in the presence of malicious clock failures (i.e. clock failures that are perceived differently by different nonfaulty clocks) is addressed. A solution to the problem, referred to as the averaging rule, is developed, and its use is analytically justified using the notions of clock partitions and generalized clock partitions. Experimental characterization of the multiple cliques problem has been undertaken, and certain conditions that induce their occurrence in practical hardware implementation are identified. The effects of clock-receiver triggering variations and phase-detector operating range on the instantaneous frequencies of the clock modules are investigated. The efficacy of the averaging rule is established not only by analysis but also by means of simulations and experimentation with hardware clock implementations. >

Derivation of Minimal Complete Sets of Test-Input Sequences Using Boolean Differences

This paper deals with a fault detection and diagnosis technique based on Boolean differences. A brief review of the notion of a Boolean difference is presented, and the concept of partial Boolean difference is introduced. An algorithm for obtaining minimal, complete sets of test-input sequences based on the partial Boolean differences of a switching function is formulated, and illustrations demonstrating the use of the technique are presented.

Effect of repair policies on software reliability

Software reliability is an important metric that quantifies the quality of the software product and is inversely related to the number of unrepaired faults in the system. Fault removal is a critical process in achieving the desired level of quality before software deployment in the field. Conventional software reliability models assume that the time to remove a fault is negligible and that the repair process is perfect. We examine various kinds of repair scenarios, and analyze the effect of these fault removal policies on the residual number of faults at the end of the testing process, using a non-homogeneous continuous time Markov chain. The fault removal rate is initially assumed to be constant, and it is subsequently extended to cover time and state dependencies. These fault removal scenarios can be easily incorporated using the state space view of the non-homogeneous Poisson process.

A non-homogeneous Markov software reliability model with imperfect repair

This paper reviews existing non-homogeneous Poisson process (NHPP) software reliability models and their limitations, and proposes a more powerful non-homogeneous Markov model for the fault detection/removal problem. In addition, this non-homogeneous Markov model allows for the possibility of a finite time to repair a fault and for imperfections in the repair process. The proposed scheme provides the basis for decision making both during the testing and the operational phase of the software product. Software behavior in the operational phase and the development test phase are related and the release time formulae are derived. Illustrations of the proposed model are provided.

Design of fault-tolerant clocks with realistic failure assumptions

The authors address the problem of designing fault-tolerant, phase-locked clocks in the presence of different types of clock failures and show that significant improvements in hardware complexity and reliability can be achieved when failed clock modules are partitioned into two classes: malicious and nonmalicious. They show that the condition N>2t+max(t1, 1) is necessary and sufficient to tolerate up to t failed clock modules out of which a maximum of t1 can behave maliciously. The practical value of this design concept is demonstrated by examples.<<ETX>>

Testing with correlated test vectors

The authors present a new built-in self-test (BIST) strategy based on correlated test vectors produced by a weighted random test-pattern generator. It is demonstrated that use of correlated test vectors reduces greatly the hardware complexity of the generator without causing significant degradation in the test outcome. Performance evaluation of this BIST technique is carried out quantitatively via probabilistic methods, and experimentally through deterministic fault simulation. Correlated weighted test patterns using the proposed scheme were applied to the ISCAS 1985 benchmark circuits, and comparisons are made to test results obtained using such techniques as WARP and GSCAN.<<ETX>>

Design of a bus-monitor for real-time applications

Abstract This paper presents the design of a bus-monitor unit whose function is to detect errors in triple modular redundant systems by monitoring their respective buses (i.e., data, address, and control buses) in real time. The bus-monitor unit is also capable of isolating single module failures in the redundant system configuration, and appropriately reconfiguring the fault-free system buses in order to overcome the deleterious effects of a failure. Although the errors considered can be the manifestation of either transient or permanent faults, the bus-monitor unit initially assumes all errors to be manifestation of transient faults, deferring their final classification and reporting until sufficient observation and evaluation of error activity has been made.