IEEE SENSORS JOURNAL, VOL. 16, NO. 22, NOVEMBER 15, 2016 8027 Chaos Time Domain Reflectometry for Online Defect Detection in Noisy Wired Networks Fabrice Auzanneau, Nicolas Ravot, and Luca Incarbone Abstract—In many application domains, wire faults can have dramatic consequences. Live wire diagnosis is often required to ensure permanent monitoring of the health of embedded cables. A novel reflectometry-based method for online wire diagnosis is presented. Chaos time domain reflectometry (CTDR) takes benefit of the properties of chaotic signals and shows very good potential for the diagnosis of live wires (i.e., during their operational usage) and complex topology networks. In particular, CTDR shows high performances in very noisy environments: the detection and the location of hard defects are possible even in the case of negative signal to noise ratio and if several reflectometers inject their signals in the cable. This enables using CTDR for the distributed diagnosis of live complex topology networks of lengths up to several tens of meters. CTDR’s defect detection capacity is shown and experimentally verified: increasing the length of the probe signal lowers the noise level. A noise robustness analysis provides a means to choose the signals parameters necessary to ensure specified detection performances. Index Terms— Reflectometry, chaos, cable, diagnosis, network. I. I NTRODUCTION E VEN in our more and more wireless world, most systems still rely on cables for energy and information transfer and use sometimes very long cables that can be subject to aggressive environmental conditions [1]. The need for diagnosis methods and systems able to quickly detect and accurately locate faults in complex wired networks has arisen around the year 2000, specifically for aeronautics [2] and energy distribution (Smart Grids) applications. Among many interesting methods, reflectometry-based methods have proven to be the most efficient ones. Reflec- tometry is used in many application domains, from humidity measurement in soils [3] to conductivity measurement in semiconductors [4]. Similarly to Radar, reflectometry injects a probe signal at one end of the cable under test. This signal propagates along the cable and each impedance discontinuity met (connector, junction, load or defect [5]) reflects a part of its energy back to the injection port. The analysis of the received signal provides information on the presence, the loca- tion and the type of these discontinuities [6]. This information is very rich and valuable for maintenance operations and health monitoring of electrical networks. Manuscript received July 22, 2016; revised August 8, 2016; accepted August 30, 2016. Date of publication September 7, 2016; date of current version October 13, 2016. The associate editor coordinating the review of this paper and approving it for publication was Prof. Danilo Demarchi. F. Auzanneau and N. Ravot are with CEA, LIST, F-91191 Gif-sur-Yvette, France (e-mail: fabrice.auzanneau@cea.fr; nicolas.ravot@cea.fr). L. Incarbone is with Sonceboz Automotive SA, CH-2605 Sonceboz, Switzerland. Digital Object Identifier 10.1109/JSEN.2016.2606567 The diagnosis of complex topology networks may require the use of several diagnosis systems in parallel [7], each one providing a different view of the network, thus eliminating location ambiguities. This has been made possible by the adoption of spread spectrum techniques [8] enabling sev- eral reflectometers to work concurrently without interfering. For safety critical systems, it may be necessary to monitor the health of the cables during their normal operation. Online diagnosis is then used, requiring harmlessness: diagnosis signals must not interfere with communication signals, and useful signals must not trigger any false alarm. Specific methods have been designed for this, such as Multicarrier Reflectometry [9], Multi Carrier Time Domain Reflectome- try (MCTDR) [10], [11] and Orthogonal Multi-Tone Time Domain Reflectometry (OMTDR) [12], mostly in the objective of diagnosis spectrum control. All these reflectometry-based wire diagnosis methods have shown to be very efficient for the detection and location of hard defects [13], i.e. defects which prevent energy from going any further (open and short circuits), and always in the case of stationary defects (meaning that the duration of the defect is longer than that of the reflectometry probe signal) [14]. This paper investigates a new reflectometry-based method, called Chaos Time Domain Reflectometry (CTDR) [15], [16], suited for the detection of hard and soft defects in wires, taking advantage of chaos theory properties. Section II recalls the bases of chaos and reflectometry and presents the main specificities of CTDR. Section III applies this method to the distributed diagnosis of defects in live networks and shows its robustness to noise, and section IV provides experimental results. Section V concludes the paper. II. CHAOS-BASED REFLECTOMETRY A. Chaos Theory Chaos theory describes systems whose behavior is highly dependent on initial conditions, and becomes very difficult to predict in time. Meteorologist Edward Lorenz was one of the first to see this phenomenon in the early 60s [17]. Often called “butterfly effect” it is described as follows: “A flutter of butterfly wings in Brazil is likely to later trigger a tornado in Texas”. The very low disturbance created by the flight of a butterfly could in principle vary the initial conditions of an atmospheric system and cause climate change later in any part of our planet. Since then, chaos theory has grown and been applied to many domains, from astrophysics to economy, computer 1558-1748 © 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.