Residential links under the weather

Abstract

Weather is a leading threat to the stability of our vital infrastructure. Last-mile Internet is no exception. Yet, unlike other vital infrastructure, weather's effect on last-mile Internet outages is not well understood. This work is the first attempt to quantify the effect of weather on residential outages. Investigating outages in residential networks due to weather is challenging because residential Internet is heterogeneous: there are different media types, different protocols, and different providers, in varying contexts of different local climate and geography. Sensitivity to these different factors leads to narrow categories when estimating how weather affects these different links. To address these issues we perform a large-scale study looking at eight years of active outage measurements that were collected across the bulk of the last mile Internet infrastructure in the United States.

Longitudinal Study of an IP Geolocation Database

“…We manually extract from each paper the granularity required (country, city, or AS) and the type of IP addresses geolocated (all, end users, end host infrastructure, and router). The "end user" category contains IP addresses belonging to residential users (e.g., [25]), or, more broadly, end users issuing web traffic (e.g., [27]). The "end host infrastructure" category includes addresses belonging to Internet infrastructure, typically web [12], proxies [37], or DNS [30] servers.…”

KL-Dection: An Approach to Detect Network Outages Based on Key Links

“…Other outage detection approaches detected network outages by probing all links in the detection area and analyzing the number of probe responses [5,7,19,20]. Heidemann et al [20] and Dainotti et al [19] used pings to probe all links in detection areas and detected the network outages by observing the apparent decrease in the number of probe responses.…”

“…However, [5] did not study the effect of weather on last-mile Internet performance, and the performance of last-mile networks affects the network connectivity of a large number of users. Hence, Padmanabhan et al [7] used ThunderPing [21] to probe all residential links in the detection area and obtained the number of probe responses.…”

“…In addition, we also give the distribution of probedistance centrality for all links according to Equation (6) and Equation (7). In Figure 2(c), the value of probedistance centrality is divided into six bins.…”

“…Quan et al detected the network outages by deploying the probes to observe the connectivity of all links in edge networks [5]. Padmanabhan et al detected the network outages by using ThunderPing [7] to measure the connectivity of all residential links in detection areas [7].…”