Distributed intelligent traffic informatics using fiber sensing
Aspects of the present disclosure describe systems, methods and structures providing wide-area traffic monitoring based on distributed fiber-optic sensing (DFOS) that employs deep neural network(s) for denoising noisy waterfall traces measured by the DFOS. Such systems, methods, and structures according to aspects of the present disclosure may advantageously monitor multiple highways/roadways using a single interrogator and optical fiber switch(es) which provides traffic information along every sensing point of existing, deployed, in-service optical telecommunications facilities.
1. A distributed traffic informatics system comprising:
a plurality of lengths of optical fiber, each individual length of said plurality of lengths of optical fibers respectively positioned alongside an individual roadway route of a plurality of roadway routes supporting vehicular traffic;
an optical interrogator unit that generates optical pulses, introduces them into the optical fiber, and receives Rayleigh backscattered signals from the individual lengths of the plurality of lengths of optical fiber;
an optical switch interposed between the interrogator and the individual lengths of the plurality of lengths of optical fiber providing selective optical communication between the interrogator and the individual lengths of optical fiber of the plurality of lengths of optical fiber, and
a data processor unit that is configured to:
determine from the backscattered signals, mechanical vibrations experienced by each of the individual lengths of optical fiber of the plurality of optical fibers resulting from a vehicle operating on the individual roadway routes alongside the individual lengths of optical fiber of the plurality of optical fibers; and
determine characteristics of the vehicle that produced the determined mechanical vibrations and the individual roadway route on which the vehicle operated.
2. The system of claim 1 further comprising:
a vehicle classifier and weight including a deep neural network for weight-in-motion (WIM) applications.
3. The system of claim 2 wherein the data processor is configured to generate from the backscattered signals time-distance (2-dimensional waterfall) graphs (images) representative of the vibrations experienced by the optical fiber along its length.
4. The system of claim 3 wherein the data processor is configured to normalize the time-distance graphs through the effect of a column normalization technique where each column of the time-distance graph represents vibration data collected at a particular location along the length of the optical fiber.
5. The system of claim 4 wherein the data processor is configured to perform the column normalization technique such that a sum of values in each column over each one minute duration is set to one.
6. The system of claim 5 wherein the data processor is configured to denoise the time-distance graphs.
7. The system of claim 3 wherein the classifier segments each pixel comprising the time-distance graphs into one of two classes selected from the group consisting of presence of a vehicle, and absence of a vehicle.
8. The system of claim 2 wherein the neural network is trained on synthetic, noisy, two dimensional time-distance data.
9. The system of claim 1 wherein the plurality of lengths of optical fiber carry telecommunications traffic.