Traditional IP routing

Routing is the primary function of IP. IP datagrams are processed and forwarded by routers which relay traffic through paths set up by various routing protocols. Routing in todays fixed networks is based on network aggregation combined with best matching. TCP/IP hosts use a routing table to maintain knowledge about other IP networks and IP hosts. Networks are identified by using an IP address and a subnet mask, and routes to single hosts are rarely set up. When a packet is to be forwarded, the routing table is consulted and the packet is transmitted on the interface registered with a route containing the best match for the destination. If no network matches are found, a default route is used if one exists.

When configuring a network interface with an IP address, a route to the network the address is a member of is usually registered on the interface automatically. This route is not set up with a gateway(the next hop along the path to the host) since hosts with addresses within this network are assumed to be reachable directly from this interface. This shows that the traditional IP routing maintains an idea of all hosts within the same subnet being on the same link. This means that all hosts in a subnet are available on a single one-hop network segment, typically via routers or switches. When working on wireless multi-hop networks this is not the case. One needs to redefine the idea of nodes being available ``on the link''. In MANETs nodes routes traffic by retransmitting packets on the interface it arrived. This approach breaks with the wired ``on-link'' way of thinking.

MANET requires a different mindset when it comes to routing. Aggregation is not used in MANETs, all routing is host based. This means that for all destinations within the MANET, a sender has a specific route. In a wired network this is not necessary due to the fact that all nodes in the local network are considered available on the link.