This paper presents a fault-tolerant object group (FTOG) framework that provides both group management service and load scheduling service for consistency maintenance as well as state transparency. The group management service manages the overall FTOG framework while providing a couple of schemes to overcome unexpected faults. The load scheduling service effectively distributes the client requests to improve the quality of service by considering the load of service objects. We develop an intelligent home network simulator (IHNS) as a simulation program for using the FTOG framework on distributed environments. Through certain simulations, we validate that the FTOG framework supports both of reliability and stability of the distributed home-networking systems

Color segmentation takes a great attention because color is an effective and robust visual cue for characterizing an object from the others. However, color segmentation suffers from color variations incurred by irregular illumination changes. We propose a reliable color modeling approach in HSI (hue-saturation-intensity) color space while considering intensity information by adopting the B-spline curve fitting to make a mathematical model for statistical characteristics of a color with respect to intensity. It is based on the fact that color distribution of a single-colored object is not invariant with respect to brightness variations even in the HS (hue-saturation) plane. The statistical characteristics contain the mean and standard deviation of hue and saturation with respect to intensity. They are mathematically expressed as four bar graphs. In order to fit the bar graphs to continuous curves, we use B-spline curve fitting procedure. From several experimental results, we verify that the proposed algorithm is successfully applied to color segmentation under various illumination conditions

This paper proposes a novel and efficient global intelligent digital redesign technique for a Takagi-Sugeno (T-S) fuzzy system. The term of intelligent digital redesign involves converting an existing analog fuzzy-model-based controller into an equivalent digital counterpart in the sense of state-matching. The proposed method should be notably discriminated from the previous works in that it allows us to globally match the states of the overall closed-loop T-S fuzzy system with the predesigned analog fuzzy-model-based controller and those with the digitally redesigned fuzzy-model-based controller, and further to examine the stabilizability by the redesigned controller in the sense of Lyapunov. The key idea is that the global intelligent digital redesign problem is viewed as a convex minimization problem of the norm distance between nonlinearly interpolated linear operators to be matched. Sufficient conditions for the global state-matching and the stability of the digitally controlled system are formulated in terms of linear matrix inequalities (LMIs). A complex nonlinear system, Duffing-like chaotic oscillator is simulated and demonstrated to validate the feasibility and effectiveness of the proposed digital redesign technique, which implies the safe applicability to the digital control system.

The authors propose a new learning algorithm for multilayer feedforward neural networks, which converges faster and achieves a better classification accuracy than the conventional backpropagation learning algorithm for pattern classification. In the conventional backpropagation learning algorithm, weights are adjusted to reduce the error or cost function that reflects the differences between the computed and the desired outputs. In the proposed learning algorithm, the authors view each term of the output layer as a function of weights and adjust the weights directly so that the output neurons produce the desired outputs. Experiments with remotely sensed data show the proposed algorithm consistently performs better than the conventional backpropagation learning algorithm in terms of classification accuracy and convergence speed.

Information on Fault Tolerance Latency (FTL), which is defined as the total time required by all sequential steps taken to recover from an error, is important to the design and evaluation of fault-tolerant computers used in safety-critical real-time control systems with deadline information. In this paper, we evaluate FTL in terms of several random and deterministic variables accounting for fault behaviors and/or the capability and performance of error-handling mechanisms, while considering various fault tolerance mechanisms based on the trade-off between temporal and spatial redundancy, and use the evaluated FTL to check if an error-handling policy can meet the Control System Deadline (CSD) for a given real-time application.

An instruction-retry policy is proposed to enhance the fault-tolerance of triple modular redundant (TMR) controller computers by adding time redundancy to them. A TMR failure is said to occur if a TMR system fails to establish a majority among its modules' outputs due to multiple faulty modules or a faulty voter. Either multiple consecutive TMR failures the active period of which exceeds a certain time limit or the exhaustion of spares as a result of frequent system reconfigurations may result in failure to meet the timing constraints of one or more tasks, called the dynamic failure, during a given mission. An optimal instruction-retry period is derived by minimizing the probability of dynamic failure upon detection of either a masked (by the TMR) error or a TMR failure. We also derive the minimum number of spares needed to keep below the pre-specified level the probability of dynamic failure for a given mission by using the derived optimal retry period.

Although Triple Modular Redundancy (TMR) has been widely used to mask the effects of a single faulty module, it cannot tolerate coincident faults in multiple modules caused by a common source, such as an environmental disruption or malfunction of a shared component. We propose a method to eliminate or alleviate the effects of (near) coincident faults by sequencing tasks on different modules in a TMR system. Specifically, we develop an effective sequencing of tasks to simply place an "optimal" distance (in the sense of minimizing the mean number of faulty tasks due to TMR failures) between the copies of a task to be executed on different modules. Several examples are presented, showing significant improvements in reducing TMR failures with the proposed task sequencing.

Failure to establish a majority among the processing modules in a triple modular redundant (TMR) system, called a TMR failure, is detected by using two voters and a disagreement detector. Assuming that no more than one module becomes permanently faulty during the execution of a task, Re-execution of the task on the Same HardWare (RSHW) upon detection of a TMR failure becomes a cost-effective recovery method, because 1) the TMR system can mask the effects of one faulty module while RSHW can recover from nonpermanent faults, and 2) system reconfiguration-Replace the faulty HardWare, reload, and Restart (RHWR)-is expensive both in time and hardware. We propose an adaptive recovery method for TMR failures by "optimally" choosing either RSHW or RHWR based on the estimation of the costs involved. We apply the Bayes theorem to update the likelihoods of all possible states in the TMR system with each voting result. Upon detection of a TMR failure, the expected cost of RSHW is derived with these likelihoods and then compared with that of RHWR. RSHW will continue either until it recovers from the TMR failure or until the expected cost of RSHW becomes larger than that of RHWR. As the number of unsuccessful RSHW's increases, the probability of permanent fault(s) having caused the TMR failure will increase, which will, in turn, increase the cost of RSHW. Our simulation results show that the proposed method outperforms the conventional reconfiguration method using only RHWR under various conditions.>

A method for calculating the hard deadlines in linear time-invariant control systems by considering system stability and the allowed state-space is presented. To derive necessary conditions for asymptotic system stability, the state difference equation is modified based on an assumed maximum delay and the probability distribution of delays whose magnitudes are less than, or equal to, the assumed maximum delay. The allowed state-space-which is derived from given input and state constraints-is used to calculate the hard deadline as a function of time and the system state. A one-shot delay model in which a single event causes a dynamic failure is considered. The knowledge of hard deadline is applied to the design of error recovery in a triple modular redundant controller computer.>