Replacement Policies for a Unit with Random and Wearout Failures

This paper considers three replacement models with random and wearout failures; a) the unit is replaced at failure, b) the unit undergoes minimal repair at failure, and c) the unit is replaced at failure only in a wearout failure period. Optimum replacement policies which minimize the s-expected cost rate for each model are discussed.     

Implications of accelerated self-healing as a key design knob for cross-layer resilience

In this paper we propose a cross-layer accelerated self-healing (CLASH) system which “repairs” its wearout issues in a physical sense through accelerated and active recovery, by which wearout can be reversed while actively applying several accelerated self-healing techniques, such as high temperature and negative voltages. Different from previous solutions of coping with wearout issues (e.g. BTI) by “tolerating”, “slowing down” or “compensating”, which still leave the irreversible (permanent) wearout component unchecked, the proposed solution is able to fully avoid the irreversible wearout through periodic rejuvenation, and this is inspired by the explored frequency dependent behaviors of wearout and (accelerated and active) recovery based on measurements on FPGAs. We demonstrate a case where the chip can always be brought back to the fresh status by employing a pattern of 31-h regular operation (under room temperature and nominal voltage) followed by a 1-h accelerated self-healing (under high temperature and negative voltage). The proposed system integrates the notions of accelerated self-healing across multiple layers of the system stack. At the circuit level, a negative voltage generator and heating elements are designed and implemented; at the architecture level, the core can be allocated in a way such that the dark silicon or redundant resources can be healed by active elements; at the system level, right balance of stress and accelerated/active recovery can be employed by the system scheduler to fully mitigate the wearout; various wearout sensors act as the media between different layers. Overall, these techniques work together to guarantee that the whole system performs for more of the time at higher levels of performance and power efficiency by fully taking advantage of the extra opportunities enabled by the accelerated self-healing.     

Time-shifted rayleigh density for wearout failures

A time-shifted Rayleigh density model is proposed for wearout failures in reliability theory. Reliability, failure rate and MTTF in the wearout period are calculated. The model predicts a linearly increasing failure rate due to wearout after the useful life.     

Analysis of errors in estimating wearout characteristics of time-dependent dielectric breakdown using system-level accelerated life test

Reliability issues exacerbated by small feature sizes in modern VLSI circuits challenge an accurate reliability assessment using the conventional approach of employing device-level accelerated life test. Since such device-level reliability assessment ignores tolerance of a circuit or a system to device wearout failures, to accurately estimate circuit/system reliability, we need to directly test a circuit or a system for extraction of wearout parameters in operating environments. In this paper, we propose a system-level accelerated life test to compliment device-level accelerated life test. We also investigate errors in estimating wearout parameters from time-dependent dielectric breakdown (TDDB) from experimental results from system-level accelerated life test and note differences from device-level reliability assessment.     

Characterizing wearout, breakdown and trap generation in thin silicon oxide

A model describing how wearout leads to breakdown in thin silicon oxides has been developed. During wearout defects or traps are generated inside the oxide and at the oxide interfaces. The signature of the trap generation is the permanent change in the transient current, in response to a voltage pulse, from an exponential decay to a 1/time decay. In oxides thinner than approximately 20 nm the dominant trap generation mechanism appears to be determined by the high fields across the oxides and not electron flow through the oxides. Locally higher current densities, flowing through the traps generated during wearout, lead to local breakdown. This model is critically dependent on the measurement of the properties of the traps generated inside the oxides during the wearout phase. The techniques for measurement of these traps and some of their properties have been described. The ability of this model to describe oxide charging, low-level leakages, transient currents, the role of asperities, polarity dependences, and the fluence, time, thickness, voltage and temperature dependences of oxide breakdown distributions has been discussed.     

Evaluation of Repairable System Reliability Using the ``Bad-As-Old' Concept

It is usually assumed that the underlying distribution of times to failure of systems is the exponential distribution. This is justified on the basis of the bathtub curve or Drenick's theorem, but the bathtub curve is merely a statement of plausibility and conflicts with Drenick's theorem. Even if exponentiality is not assumed, it is usually assumed that a system under study is as-good-as-new after repair. This is not a plausible assumption to make for a complex system. If failure data are available they should be tested for trend among successive failure times. If a trend exists, a time dependent (nonhomogeneous) Poisson process (called bad-as-old model in this paper) should be fitted and tested for adequacy. This paper is not intended to provide a rigorous, definitive treatment of bad-as-old models. Rather, it has three main purposes: 1) to point out the glaring, but somehow usually overlooked, inconsistency between the commonly accepted concept of wearout of repairable systems and the a priori use of renewal processes for modeling these systems; 2) to outline basic procedures for evaluating data from repairable systems and for formulating bad-as-old probabilistic models; and 3) to present the results of Monte Carlo simulations, which illustrate the grossly misleading results which can occur if independence of successive failure times is invalidly assumed.     

Checking request policies for a one-unit system and their comparisons

In a checking problem, a system is not necessarily checked without delay at checking time, i.e., the inspector who performs its checking is not necessarily in standby costantly. We consider a one-unit system in which the inspector is requested to check the condition of the system and the checking by him is performed after his arrival at the system. We define such a checking policy as a checking request policy. Applying the unique modifications of Markov renewal processes, we analyze each stochastic behavior of the models for checking request policies I and II. The optimum checking request policies maximizing the cost effectiveness are derived under suitable conditions for each model. We compare these two checking request policies and give selecting criteria between checking request policies I and II. Finally, we show numerical examples on the selecting criteria.     

Periodic-replacement models with threshold levels

The authors suggest five replacement policies where a unit is replaced at periodic times, jT(j=1,2, . . .), and the replacement cost is expensive when some number of events occurring in (0,t) is greater than a threshold level. The usual models for inspection, periodic replacement, block replacement, parallel systems, and cumulative damage can be transformed into replacement models with threshold levels. The mean cost-rate of each model is obtained, using well-known results of reliability theory. The optimum replacement time which minimizes the cost-rate of an inspection model is discussed and shown to exist uniquely     

The Analysis of a Statistical Multiplexer with Nonindependent Arrivals and Errors

Most stochastic models for discrete time statistical multiplexers make the assumption that 1) arrivals are governed by a Bernoulli process and that 2) transmission errors or multiplexer failures are described by an independent error process. We study a multiplexer operating in a two state Markovian environment in which each state is characterized by its own Bernoulli arrival process and independent error process. We derive the probability generating function for the queue length distribution for such a system. We also consider two special cases of this model. One referred to as the saturated arrival model corresponds to a system in which during one of the two states, the saturated state, at least one arrival occurs during each discrete time unit. The other model, the breakdown model, corresponds to a system in which during one of the two states, the breakdown state, the multiplexer is inoperative. For both models, we generalize the analysis to cases in which the durations of the saturated state and breakdown state may take on values described by arbitrary distributions. Finally, we study the effects of different arrival processes and error processes on queue length behavior.     

Failure mechanism models for cyclic fatigue

This work illustrates design situations where mechanical fatigue under cyclic loading, of one or more components, can compromise system performance. In this failure mechanism, damage accumulates with each load cycle, thereby causing a physical wearout failure mechanism. Phenomenological continuum length-scale models, based on micromechanical considerations, are presented to predict the onset (or initiation) of fatigue cracking in ductile materials. Fatigue crack propagation is modeled with continuum fracture mechanics principles. The number of load cycles required to cause failure is predicted based on these models. Approaches for modeling creep fatigue interactions are briefly discussed. Analytic physics-of-failure method and examples are presented for designing against wearout failure due to cyclic fatigue. These models can be implemented in an engineering design environment. The associated stress analysis requires numerical finite element techniques in many cases. The associated material property characterization techniques have matured since the 1950s and are specified in engineering handbooks     

Failure mechanism models for material aging due to interdiffusion

This tutorial illustrates design situations where material aging due to interdiffusion in some components can compromise system performance over time, thereby acting as a wearout failure mechanism. Microstructural diffusion mechanisms, continuum diffusion models, and interdiffusion analysis techniques are presented to design against such failures. An example illustrates the application of the mechanisms, models, and techniques in electronic packaging     

Genetic algorithms applied to cellular call admission: localpolicies

It is well known that if a stochastic service system (such as a cellular network) is shared by users with different characteristics (such as differing handoff rates or call holding times), the overall system performance can be improved by denial of service requests even when the success capacity exists. Such selective denial of service based on the system state is defined as the call admission. A previous paper suggested the use of genetic algorithms (GAs) to find near-optimal call admission policies for cellular networks. In this paper, we define local call admission policies that make admission decisions based on partial state information. We search for the best local call admission policies for one-dimensional (1-D) cellular networks using genetic algorithms and show that the performance of the best local policies is comparable to optima for small systems. We test our algorithm on larger systems and show that the local policies found outperform the maximum packing and best handoff reservation policies for the systems we have considered. We find that the local policies suggested by the genetic algorithm search in these cases are double threshold policies. We then find the best double threshold policies by exhaustive search for both 1-D and Manhattan model cellular networks and show that they almost always outperform the best trunk reservation policies for these systems     

Improved cost effectiveness for new satellite systems

Concepts of reliability, life and earning power are applied in a new way to measure the cost effectiveness and depreciation of satellite systems in view of requirement versus capability.The operational reliability of Telesat's present satellite systems is evaluated against the early assumptions of constant failure rates. New realistic reliability wearout models are developed and result in reliability predictions that correlate well with operational data.Optimization and selection of cost effective configurations for new increased capability satellite systems are achieved by a series of tradeoffs using the new reliability wearout models. The new systems show a substantial improvement in the level of cost effectiveness optimization over a substantially longer life in comparison with the present systems. The depreciation of the new systems is also shown to be more uniformly distributed over longer life.     

Generalized multistate monotone coherent systems

Most available models for multistate coherent systems (MCS) assume that the state sets of the system and its components are totally ordered-an assumption that greatly limits the application of the usual MCS models. This paper presents a new MCS model-a generalized multistate coherent system (GMCS) model-which assumes, more generally, that the state sets of the system and its components are partially ordered. Since the structure of a partially ordered set is very flexible, the model applies to the reliability analysis of various multistate coherent systems. As a result, the GMCS model generalizes the coherent system theory from the binary-state case to the multistate case. The authors investigate some of the properties of this generalized model. Then, system reliability is redefined and bounds for its evaluation are derived     

Performance analysis of power management policies in wireless networks

It has long been recognized that energy conservation usually comes at the cost of degraded performance such as longer delay and lower throughput in stand-alone systems and communication networks. However, there have been very few research efforts in quantifying such trade-offs. In this paper, we develop analytical models to characterize the relationships among energy, delay and throughput for different power management policies in wireless communication. Based on the decision when to put nodes to low-power states, we divide power management policies into two categories, i.e., 1) time-out driven and 2) polling-based. M/G/1/K queues with multiple vacations and an attention span are used to model time-out driven policies while transient analysis is applied to derive the state transition probability in polling-based systems. We find that For time-out driven power management policies, the "optimal" policy exhibits a threshold structure, i.e., when the traffic load is below certain threshold, a node should switch to the low-power state whenever possible and always remain active otherwise. From our analysis, contrary to general beliefs, polling-based policies such as the IEEE 802.11 PSM are not energy efficient for light traffic load.     

Failure-mechanism models for creep and creep rupture

This tutorial illustrates design situations where creep and creep rupture of components can compromise system performance over time, thereby acting as a wearout failure mechanism. Polycrystalline materials, such as metals and ceramics, and polymers are treated. Analytic microstructural creep mechanisms leading to failure of these materials are presented. Continuum microscale models for predicting long-term creep are explained for practical design purposes. Two examples illustrate these models for mechanical engineering and electronic packaging situations     

Abrupt breakdown in dielectric/metal gate stacks: a potential reliability limitation?

In downscaled poly-Si gate MOSFET devices reliability margin is gained by progressive wearout. When the poly-Si gate is replaced with a metal gate, the slow wearout phase observed in ultrathin SiON and HfSiON dielectrics with poly-Si gate disappears, and with it, the reliability margin. We demonstrate for several combinations of dielectric and gate materials that the large abrupt current increase (/spl Delta/I) as compared to poly-Si is not likely due to process issues, but is an intrinsic property of the dielectric/metal gate stack. The occurrence of large /spl Delta/I is a potential limitation for the reliability of metal gate devices.     

Reliability analysis of a 2-out-of-n:F system with repairable primary and degradation units

In this paper we consider a 2-out-of-n:F system composed of repairable primary and degradation units. Two stochastic models are proposed depending upon the policies of the repair discipline. We derive the availability, the expected number of visits to system failure per unit time and the MTBF by applying an extended Markov renewal process for each model. We finally show the numerical examples and investigate the impacts of introducing degradation units on the reliability measures obtained above.     

Optimal maintenance policies under different operational schedules

In the reliability literature, maintenance time is usually ignored during the optimization of maintenance policies. In some scenarios, costs due to system failures may vary with time, and the ignorance of maintenance time will lead to unrealistic results. This paper develops maintenance policies for such situations where the system under study operates iteratively at two successive states: up or down. The costs due to system failure at the up state consist of both business losses & maintenance costs, whereas those at the down state only include maintenance costs. We consider three models: Model A, B, and C: /spl middot/ Model A makes only corrective maintenance (CM). /spl middot/ Model B performs imperfect preventive maintenance (PM) sequentially, and CM. /spl middot/ Model C executes PM periodically, and CM; this PM can restore the system as good as the state just after the latest CM. The CM in this paper is imperfect repair. Finally, the impact of these maintenance policies is illustrated through numerical examples.