Adaptive routing is the process of locating a clear path from a source point to a destination point across a network of nodes that could change at any point. In computer networking, adaptive routing algorithms are used to ensure that data packets can move from one point in the network to another, even if the one or more nodes in between are unavailable. The concepts of adaptive routing are also used by software to provide travel instructions in global positioning systems and can be used in other applications, such as artificial intelligence.
The most widely available example of adaptive routing is the Internet, which is essentially a worldwide computer network. Data travels across this network from node to node, server to server. The Internet is a dynamic network and implements special routing algorithms. This means that if several servers between two cities are unavailable, the algorithms will find a new path from one city to the other through whatever nodes are available.
Networks that do not employ adaptive routing are known as fixed routing networks. These networks have set paths that packets move through with no ability to change the routes unless an administrator reprograms or rewires the operational routers. Fixed routing networks have the advantage of sometimes being faster over certain distances, but they have the vulnerability of completely failing in packet delivery if a vital node is compromised.
Even in a dynamically routed network, the most direct and shortest path is first attempted. Using protocols, data is able to collect the topography of the network it is moving across to determine what routes are viable and which route is the quickest. This information relies on the state of a network. State information is transferred through adaptive routing protocols from each node.
Several adaptive routing protocols have been developed for different purposes. Intermediate-system-to-intermediate-system (IS-IS) protocol is designed to route data through large networks like Internet backbones. The routing information protocol (RIP) is excellent for small-distance transport. Some experimental protocols attempt to increase the speed and efficiency of the movement of packets without requiring full knowledge of the network state.
The same concepts that are used to find the path from one node in a network to another can also be applied to applications such as global positioning systems. With information about the status of a city and its streets, an effective route from one destination to another can be plotted. The addition of real-time traffic information can add a weight to each node, or street, in a city, allowing the fastest course can be determined.