One number no consumer ever sees advertised is "concentration" ratio, though every Internet service provider makes deliberate decisions about how much to "oversubscribe" its bandwidth. A typical cable modem or digital subscriber line service aimed at consumer users might routinely use a 200:1 contention ratio, where 10 Mbps of Internet bandwidth is shared by 200 subscribers, on the assumption that they won't all be online at the same time.

 

T1 services more generally aimed at business customers typically use a less-severe concentration ratio of perhaps 8:1.

 

The reason that is significant is that the contention ratio determines effective throughput at peak hours of usage. All core bandwidth is shared in the access network. The issue is how much sharing, and where the sharing occurs. That is one good reason why the effective bandwidth any single user might experience is only partially determined by the stated peak bandwidth of an access circuit.

 

Contention is an issue for both wireless and wired access networks, and though radio capacity obviously is shared by everybody trying to use a single cell site, so is capacity for all cable modem users in a single neighborhood served by an optical node, as well as bandwidth shared by all DSL customers on the "trunk" side of a digital subscriber line access mulitiplexer.

 

Fiber-to-home services such as Verizon's FiOS (News - Alert) also use concentration, so the issue in some cases is how to compare the "real world" performance a wired customer might expect, compared to similar performance using a 3G or 4G wireless connection, assuming similar customer penetration (obviously a lightly-loaded network is going to perform better than a heavily-loaded network, and 4G networks will have light loading at first).

 

Still, as Long Term Evolution networks progress from scores of megabits per second up to about 1 Gbps, and as fiber-to-curb networks offer 20 Mbps to 40 Mbps now, scaling to 60 Mbps or so in the future, while FiOS style networks might offer 50 Mbps, with capacity also growing to 1 Gbps in the future, the issue of effective throughput is one measure of how much subsitution potential might exist.

 

In other words, if a user can get 20 Mbps or 40 Mbps using 4G, is that a functional substitute for the same bandwidth available on a wired network? Possibly. Users who do not use lots of applications requiring video might well be able to get along quite nicely using wireless access.  

 

With moderate to heavy loading of a 4G tower, though, wired networks are certain to retain a bandwidth advantage. As a practical matter, a wired network almost has to offer more bandwidth to remain relevant. So expect effective, real-world bandwidth obtainable from wired networks to remain at a multiple of the wireless bandwidth, as wireless bandwidth continues to grow. Whether the ratio is 2:1 or 10:1 will be the issue.