Improved Algorithms for Single-Ended Loop Make-Up Identification

Digital Subscriber Lines (DSL) transmit over twisted-pair telephone loops, and knowledge of the loop make-up is vital for determining what types of DSLs may be deployed and what bit rates they may transmit. Single-ended automatic loop identification (LID) is essential for achieving low-cost deployment of DSL, since it allows loops to be qualified in bulk and does not involve any human intervention at the customer´s location. This paper describes strategies to enhance previous algorithms for loop make-up identification via single-ended measurements based on Time Domain Reflectometry. In particular, we show that clustering techniques can be effectively used to partition the set of possible loop topologies so as to reduce overall identification time. Furthermore, statistical data is used to improve overall LID accuracy by accounting for the likelihoods of different cable gauges at different distances from a central office (CO).