Talk:Network coding

From Wikipedia, the free encyclopedia

The Butterfly network is directly analogous to the connection of two telephones. The telephones have 4 wire networks (transmit and receive pairs). These are linked by a 2 wire network (a single pair) which carries the signals from both ends. Each telephone has a hybrid circuit or using another name a 2 to 4 wire converter to make the required conversions.

The use of 2 wire transmission to save on switching and transmission pairs seems to be the same reason that coding is attractive in conserving router bandwidth.

Is network coding a digital version of 2 wire (2W) transmission in telephony? It would seem to be clear that it is at a surface level. 2W wire transmission is well over a century old in telpehony.

Note that the 2 to 4W conversion analogy in the case of network coding and wireless networks.

While the statement that the Min Cut Max Flow Theorem produces the maximum theoretical capacity of a network [from the point of view of a single designated user node pair]is strictly speaking correct, it is a poor bound. This is so since its use assumes that no other nodes are inputing/outputing any data other than a specific designated pair. That is the same pair that defines the appropraite cut referred to by the Min Cut Max flow method. Consequently the use of the resultant to establish an upper bound for the capacity/utility in any useful sense is way too optimistic. Any meaningful estimate would have to assume some spatial/temporal distribution of concurrent users. Since the selection of a parametric distribution brings into play many additional degrees of modeling freedom, a prefered standard seems a long shot to say the least. Of course many more practical limitatins on the connectivity of the network [eg the reachability of a specific node from another without some necesary indirectivity] bring into play many more such model dependent conditions and parameters. Therefore it would seem that a meaningful capacity calculation based on "network topology" alone is not possible for any but the simplist "toy problems" Charmonimu5 21:58, 29 August 2007 (UTC) Richard Schmidt