We introduce an algorithm for maximizing utility through congestion control and random allocation of routes to users. The allocation is defined by a probability distribution whose degree of randomness as measured by its entropy, is controlled. We seek to use the entropy to modulate between the two extremes of path specification; minimum cost path selection on one end and equiprobable path allocation on the other. We explore the trade-off between stability and utility in two sample topologies.
Proceedings Title: Proceedings of the 5th International Conference on Cybernetics and Information Technologies, Systems and Applications: CITSA 2008
Conference Dates: June 29-July 2, 2008
Conference Location: Orlando, FL
Pub Type: Conferences
cross-layer, TCP/IP algorithms