Application of time series analysis to fault management in MANETs

Abstract

Traditional network management systems are not usually able to differentiate between mobility and other causes of communication degradation in wireless mobile ad hoc networks. A fault management system needs the ability to not only detect changes in performance but also reason about their possible causes and how to fix the problems. We propose a system called TimeSAFE (Time Series Analyzer Front End) that performs time series analysis as a front end input to a central fault management system. We show how such an analysis can help to distinguish between motion, obstacles, and interference as possible causes of changes in SINR (Signal to Interference and Noise Ratio). We propose three different methods of time series analysis and also methods for dynamic order selection and dynamic window selection. We describe our implementation of TimeSAFE and provide some preliminary results of detecting changes in SINR and identifying their causes.

Harvesting Time-Series Data from Service-Based Systems Hosted in MANETs

“…The primary evaluation question for our harvesting method is how well the dependence data of each conversation C has been propagated through the network after a certain period of time. More specifically, we measure the quality of the harvest in terms of the ratio of true positives (TP) in the dependence analysis result, defined as follows [5,6]:…”

“…To understand this effect for our case study, we use the ratio of false positives (FP) [5,6] in dependence graphs to indicate the impact of (im)precision in the dependence data, defined as follows:…”

“…Time-varying, local data are observed by monitors situated at the nodes of the network and captured as a time series. The time-series data are later harvested from the nodes and presented to a management element located somewhere in the network so that it can perform a global state analysis, such as fault identification [5,6], service discovery and composition [7], or service re-placement [8,9]. As it turns out, harvesting is an especially difficult problem in MANETs due to the inherent resource limitations of wireless devices and the fact that mobility can lead to network instabilities, asymmetries, 1 and partitions resulting from relative node movements (e.g., going in and out of range) and link properties (e.g., directionality, interference, and noise) [10].…”

Locating Faults in MANET-Hosted Software Systems

“…Vashist et al [35] use locally and globally derived symptoms of faults from statistical analysis of network traffic measurements to detect network partitioning as well as faulty links and nodes. Sapello et al [28] transfer time series data in streams from monitors to analysis nodes to detect individual faulty and noisy links. All these methods are limited by the fact that they focus on identifying the faults at low-level network components (hosts and links), not taking into account the applications actually making use of the MANET.…”

“…Existing fault localization methods designed for the MANET environment [5], [6], [8], [9], [12], [21], [22], [28] focus on network-level identification of individual faulty hosts and/or the links between them. The methods are limited to low-level observations and measurements of packet flows and host failures, and therefore are blind to the end-to-end context of the service-level conversations affected or unaffected by those faults.…”