Network Modeling and Performance Evaluation for G.fast

Abstract

G.fast is a gap-bridging broadband technology on the way to a fully optical access network. G.fast is deployed in hybrid fiber-copper access networks and aiming to offer ubiquitous low-cost and highspeed broadband. For network operators, it is crucial to determine the location from where to deploy G.fast, the expected network coverage, and the expected bit rates. In this paper, we perform network modeling and statistically assess the performance of G.fast based on actual network data in four geotype classes: urban, suburban, dense rural, and sparse rural. For each class, we have collected the network data in the field with a substantial number of twisted-pair lines in Austria. Statistical analysis of loop lengths indicates that to improve the network coverage, the G.fast should be deployed in urban and suburban areas from the socalled remote node, whereas in rural areas from the last distribution point. Under such a deployment rule, the analysis by means of empirical complementary cumulative distribution functions shows a good network coverage for all network classes. Furthermore, the simulation results indicate a significant influence of cable types. Consequently, for the benefit of the cable community, we share measurements of 695 twisted-pairs of cable types relevant for G.fast deployment commonly found in the Austrian network.