Comparison with other Ad Hoc Routing Protocols

1.21Comparison with other Ad Hoc Routing Protocols

This section provides comparisons of previously described routing algorithms (section 3) with Fisheye routing. Table 1 summarizes and compares properties of the ad hoc routing protocols.

CGSR and ZHLS differ among the other table-driven protocols, in that they use a hierarchical addressing scheme such that nodes are grouped into clusters (or zones). Nodes can be localized for channel access, routing, bandwidth allocation separation among clusters. This has the advantage that it can scale well to high network sizes. However, this relies on critical nodes to control routing between regions and to maintain node association. This is a difficult problem to solve, but may be necessary for large networks. While flat addressing schemes may be less complicated and easier to use, there are doubts as to its scalability [Dal97]. Fisheye partially circumvents the problem of scalability of flat addressing schemes by using different updating scopes. This has the effect of localizing routing messages to nodes that are close to each other.

Among the on-demand routing schemes, AODV, DSR, and TORA find shortest-hop routes only when routes to new destinations are desired. ABR and SSR are on-demand routing schemes that find routes that are longer-lived (which are not necessarily shortest hop) based on some metric. It is uncertain weather shortest hop routes or longer-lived routes are better. Since longer-lived routes do not necessarily result in smallest number of hops, it may incur higher latency. However, longer-lived routes require fewer route reconstruction and therefore may yield higher throughput. In addition, network conditions will affect the performance of each method. Long-lived routes will be favored in presence of high mobility when there are higher number of link changes, and shortest-hop routes will be favored when there is low mobility. Thus, it remains to be seen whether longer-lived routes are more optimal than shortest-hop routes.

On-demand routing schemes have an advantage over the table-driven fisheye scheme in that they do not rely on an underlying routing table update mechanism that involves the constant propagation of routing information. Routing information in Fisheye is constantly propagated, and a route to every other node in the network is available. This feature incurs substantial signaling traffic. However, in on-demand routing, routing traffic grows with increasing mobility of active routes and with increasing source/destination traffic pairs. Thus, in a large dense network with high number of traffic pairs, on-demand routing may incur higher overhead traffic than the fisheye scheme. Since network conditions are not known a priori, it is favorable to have a mechanism that is insensitive to traffic conditions.