Capability information: A cost-effective information model for multi-hop routing of wireless ad hoc networks in the real environment

摘要

In greedy routing, each relay node forwards the message to a neighbor (also called the successor) that is closer to the destination. However, the successor candidate set (SCS) is different every time the relative location of the relay node to the destination changes. The configuration in the entire network, when all succeeding paths from a relay node are blocked by local minima, is irregular and its concern region cannot be determined unless the routing actually initiates. In the real deployment environment of the wireless ad hoc networks, the link quality also changes dynamically. This brings a challenge for the local decision of the greedy advance to precisely adjust its SCS for the flip-flop of link quality that blocks the non-detour path ahead. This paper introduces a new information model to a non-detour routing, also called progressive routing, under the impact of dynamic blocks. As a result, each 1-hop advance, by sacrificing little routing flexibility, can avoid those unsafe situations and remains on a non-detour path. In our model, each node prepares the information in a proactive mode, but can use it for all different paths passing through, saving the cost and delay in the reactive mode. We focus on an "everyone" model, in which each node will apply the same generic process in a fully distributed manner, in order to achieve a scalable and reliable solution. In detail, we discuss how in a sample realistic environment the pattern of SCS can be interpreted in a single safety descriptor e [0,1] at each node. It indicates the maximum probability of a successful non-detour path from this node to the edge of networks. The larger value the more likely the non-detour routing will be successful and the more stable the path will be. We illustrate the effectiveness of this indirect reference information in the corresponding routing in terms of the success of non-detour path constitution and the ability of self-adjustment for dynamics in the networks, while the cost of information construction and update propagation is minimized. The results are compared with the best results known to date.