Browse Prior Art Database

Application of Stochastic Simulation for Real Time Estimation and Optimization of Reliability in Distributed Systems

IP.com Disclosure Number: IPCOM000236552D
Original Publication Date: 2014-May-02
Included in the Prior Art Database: 2014-May-02
Document File: 3 page(s) / 138K

Publishing Venue

Microsoft

Related People

Ruwen H Hess: INVENTOR [+2]

Abstract

The invention describes a method of providing real time estimation and optimization of reliability in a distributed system using stochastic simulation. The method (a) uses sampling of probabilistic distributions to estimate state of the components that are not monitored actively, and (b) uses component and event dependencies, measured and estimated state of components as inputs for a stochastic simulation to predict future failures of components and the overall system.

This text was extracted from a Microsoft Word document.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 53% of the total text.

Document Author (alias)

Ruwen Hess (ruwenh)

Defensive Publication Title 

Application of Stochastic Simulation for Real Time Estimation and Optimization of Reliability in Distributed Systems

Name(s) of All Contributors

Ruwen Hess

Ian Baker

 

 

 

Summary of the Defensive Publication/Abstract

The invention describes a method of providing real time estimation and optimization of reliability in a distributed system using stochastic simulation. The method (a) uses sampling of probabilistic distributions to estimate state of the components that are not monitored actively, and (b) uses component and event dependencies, measured and estimated state of components as inputs for a stochastic simulation to predict future failures of components and the overall system.

 

Suggested Keywords: Helps with Examiner Searching and Search Engine Optimization.

Stochastic Simulation, Distributed System, Real Time Estimation, Reliability Estimation/Optimization, Current/Future  State, Health Status, Prediction of Future State/Behavior/Failures, State Model, Random variables, Probabilistic Distribution, Conditional probability, Transition model

 

Description:  Include architectural diagrams and system level data flow diagrams if: 1) they have already been prepared or 2) they are needed to enable another developer to implement your defensive publication. Target 1-2 pages, and not more than 5 pages.  

Current systems rely on tracking ‘present state health’, and assuming that built in redundancy allows for enough time-buffer to address failures before they result in system outage.  At most, these systems contain an explicit but static model of the expected failure rate.

In contrast, our invention predicts the likely future behavior, taking into account dependencies between events and components, and integrating feedback from the current system into these predictions. Furthermore, our invention allows the modeling of state of components that are not actively tracked by the usual health monitoring in existing solutions.

This invention provides a way to model a distributed system and simulate component and system failure outcomes, supplying the operator with information that will allow them to prevent failures or mitigate their impact before they occur

The first step of our solution is to generate the starting state model:

As pictured in the diagram below, a number of inputs (topology details, component details including failure probability distributions, event details including conditional failure probability distributions for affected components, and operational tasks) is used to construct an initial state model (shown by (1) ). This state model describes which components are in the system, and in which topology. It also holds details for each component regarding the probabilistic distribution of their state and failure rate, as well as conditional probabilities of failure for the various possible events.

The model created is the operational model that forms the basis on wh...