User-level failure detection and auto-recovery of parallel programs in HPC systems

As the mean-time-between-failures(MTBF)continues to decline with the increasing number of components on large-scale high performance computing(HPC)systems,program failures might occur during the execution period with high probability.Ensuring successful execution of the HPC programs has become an issue that the unprivileged users should be concerned.From the user perspective,if the program failure cannot be detected and handled in time,it would waste resources and delay the progress of program execution.Unfortunately,the unprivileged users are unable to perform program state checking due to execution control by the job management system as well as the limited privilege.Currently,automated tools for supporting user-level failure detection and autorecovery of parallel programs in HPC systems are missing.This paper proposes an innovative method for the unprivileged user to achieve failure detection of job execution and automatic resubmission of failed jobs.The state checker in our method is encapsulated as an independent job to reduce interference with the user jobs.In addition,we propose a dual-checker mechanism to improve the robustness of our approach.We implement the proposed method as a tool named automatic re-launcher(ARL)and evaluate it on the Tianhe-2 system.Experiment results show that ARL can detect the execution failures effectively on Tianhe-2 system.In addition,the communication and performance overhead caused by ARL is negligible.The good scalability of ARL makes it applicable for large-scale HPC systems.

Motor speed estimation and failure detection of a small UAV using density of maxima

This work presents the application of the technique named signal analysis based on chaos using density of maxima to analyze brushless direct current motors.It uses a correlation coefficient estimated from the density of maxima of the current signal.This study demonstrates in experiments the speed estimation of a brushless motor on a testbench and failure detection in a small flying drone.The experimental results demonstrate that it is possible to estimate the speed in 97.8%of the cases and to detect failure in 82.75%of the analyzed cases.

Prediction of Link Failure in MANET-IoT Using Fuzzy Linear Regression

A Mobile Ad-hoc NETwork(MANET)contains numerous mobile nodes,and it forms a structure-less network associated with wireless links.But,the node movement is the key feature of MANETs;hence,the quick action of the nodes guides a link failure.This link failure creates more data packet drops that can cause a long time delay.As a result,measuring accurate link failure time is the key factor in the MANET.This paper presents a Fuzzy Linear Regression Method to measure Link Failure(FLRLF)and provide an optimal route in the MANET-Internet of Things(IoT).This work aims to predict link failure and improve routing efficiency in MANET.The Fuzzy Linear Regression Method(FLRM)measures the long lifespan link based on the link failure.The mobile node group is built by the Received Signal Strength(RSS).The Hill Climbing(HC)method selects the Group Leader(GL)based on node mobility,node degree and node energy.Additionally,it uses a Data Gathering node forward the infor-mation from GL to the sink node through multiple GL.The GL is identified by linking lifespan and energy using the Particle Swarm Optimization(PSO)algo-rithm.The simulation results demonstrate that the FLRLF approach increases the GL lifespan and minimizes the link failure time in the MANET.

Generalized reliability measures of Kalman filtering for precise point positioning

To deal with the adverse influence of model failures on Kalman filtering （KF） estimation, it is necessary to investigate the generalized reliability theory, including the model failure detection and identification method as well as the separability and reliability theories. Although the generalized reliability theory for the least square has been discussed for many decades, the generalized reliability theory of KF is not widely discussed. Compared with the least square, KF includes not only the measurement model, but also the dynamic model. In KF, the predicted value of the state parameters from the dynamic model is considered as pseudomeasurements and combined with the observed measurements to compose the form of the least square. According to the reliability of the least square, the generalized reliability of KF is derived. Then, the dynamic model failure of precise point positioning is simulated to demonstrate the usage of the generalized reliability theory. The results show that the adverse influence of the dynamic model failure is more severe than that of the measurement model. Moreover, it is recommended that the model failure identification should always be used even if the overall model test passes. It is shown that the derived generalized reliability measures are suitable for the generalized KF estimation.

On the Effectiveness of the System Validation Based on the Black Box Testing Methodology

With the advancement of technology in recent years, effective fault diagnosis became a necessity to verify the performance and ensure the quality of complex systems. In this paper, an original verification methodology for complex consumer electronic devices is presented. Verification of the system which consists of hardware （integrated circuit） and corresponding software within a flat panel TV set is in the focus. Proposed methodology provides reliable functional failure detection using the concept of black box testing. Further, the approach is fully automated, improving the reliability and speed of failure detection. The methodology effectiveness has been experimentally evaluated and the analysis results have been reported.

Systematic exploration of signal-based indicators for failure diagnosis in the context of cyber-physical systems

Malfunction or breakdown of certain mission critical systems(MCSs) may cause losses of life, damage the environments, and/or lead to high costs. Therefore, recognition of emerging failures and preventive maintenance are essential for reliable operation of MCSs. There is a practical approach for identifying and forecasting failures based on the indicators obtained from real life processes. We aim to develop means for performing active failure diagnosis and forecasting based on monitoring statistical changes of generic signal features in the specific operation modes of the system. In this paper, we present a new approach for identifying emerging failures based on their manifestations in system signals. Our approach benefits from the dynamic management of the system operation modes and from simultaneous processing and characterization of multiple heterogeneous signal sources. It improves the reliability of failure diagnosis and forecasting by investigating system performance in various operation modes, includes reasoning about failures and forming of failures using a failure indicator matrix which is composed of statistical deviation of signal characteristics between normal and failed operations, and implements a failure indicator concept that can be used as a plug and play failure diagnosis and failure forecasting feature of cyber-physical systems. We demonstrate that our method can automate failure diagnosis in the MCSs and lend the MCSs to the development of decision support systems for preventive maintenance.

Detecting and Locating Failures in Communication Networks

The connectivity of a strongly connected network may be destroyed after link damage.Since many net- works are connected by directed links,the reachability may be restored by altering the direction of one or more of the links and thus reconfigoring the network.The location of the failed link must first be determined.In this paper,we examine new methods to determine the location of failed links and nodes in networks.A routing test approach is proposed and the conditions under which communication networks may be tested are discussed. Finally,an adaptive algorithm and a heuristic algorithm that can locate a single failed llnk or a single failed node are presented.