A class of decentralized routing algorithms using relaxation
Abstract
An important problem in packetswitched communication networks is the optimal assignment of routes to the message packets. An optimal routing assignment is one which chooses network paths for the packets in a way that minimizes some cost function, typically average message delay. A class of optimal routing algorithms is described which utilize a type of iterative computation known as relaxation. Computation is decentralized in the sense that each node computes its routing strategy using only information supplied from adjacent nodes. Being iterative, the algorithms are inherently adaptive. The routing computation is based conceptually on an electrical network analog for the optimization problem. It is shown that a simple, convergent relaxation procedure can be used to 'solve' the analog network, thereby yielding the optimal routing strategy. A simple example is presented to illustrate the method. In general, the computational load compares favorably with other (centralized) methods, although further work is needed to obtain quantitative comparisons in specific cases.
 Publication:

IEEE Transactions on Communications
 Pub Date:
 October 1977
 Bibcode:
 1977ITCom..25.1092S
 Keywords:

 Algorithms;
 Communication Networks;
 Relaxation Method (Mathematics);
 Iteration;
 Network Analysis;
 Optimization;
 Routes;
 Communications and Radar