Network congestion issues pervade all packet network designs. Here are some DSR references: Attacks, Agorics.

There is good stuff here but it badly needs organization!

Congestions in networks, like road systems, occur on many time scales. Solutions are mostly scale dependent.

Public utilities have been traditionally managed by engineers who predict peak loads and recommend additional capacity soon enough to avoid shortages. Prices were the domain of politicians. Many have imagined alternatives. This works to varying degrees. I think that market mechanisms have largely controlled investment in long distance Internet capacity, often to the detriment of investors. The last mile and the last ten miles, however of bit transport have remained largely monopolistic, or duopolistic. DSR recommends market solutions but has no special insight on how to transcend the duopoly.

There are thus two scales that concern us regarding congestion: time and space. In the long term long distance corner of the gamut market forces will entice suppliers. This has sufficed to provide capacity but not always to reward investors.

Short timespans and long distances present a dilemma. Congestion can arise on a timescale shorter than what is necessary for a signal to transit the network. This is reminiscent of traders who had traveled from the West to the orient and were faced with making deals when they were months away from information about market prices back home. Local decisions were necessary but the packet is so far too dumb to decide, even with the beneficent help of a node. Bang-Bang Prices are the crudest stab at a solution.