Extraction of two wire and power line loop topology using customized genetic algorithms

摘要

Channel discovery by a Modem is critical to rate determination in 2-wire, smart grids and in-vehicle communications. A two-step hybrid method to extract open port 2-wire and power line topologies and facilitate channel model simulation using a biologically-inspired Genetic Algorithm (GA) iterative optimization is discussed. An initial approximate estimate of the loop topology is obtained from Correlation Time Domain Reflectometry (CTDR), a technique well-discussed in literature. Subsequently, a GA optimization working on the outputs of Frequency Domain Reflectometry (FDR) data of the loop progresses iteratively and converges to a more accurate topology. The GA defines a fitness function based on the measured FDR data and the computed FDR echo of successive loop section estimates obtained using the Roulette Wheel Selection mechanism. This was validated by actual FDR measurement tests on 2-wire loops and final topology results are compared with those obtained using an MMSE optimization scheme. For typical two-dimensional power line topologies, we present simulation results in a realistic channel in presence of impulse noise, using this two-step hybrid method. The aforementioned methods reuse most of the firmware modules in an embedded DSL modem, thus needing no additional equipment for measurements while ensuring rapid convergence.