CR: Capability Information for Routing of Wireless Ad Hoc Networks in Real Environment

Wireless ad-hoc networks (WANs) have great long-term economic potential and the ability to transform our lives. Consider the WAN application of emergent disaster recovery. Before delivering food, water, and medicine, as well as doctors to the survivors, we need to know where and how many of these things are needed. The most efficient way is to send rescue teams carrying portable equipment, to search for the victims and survivors. The environment information will be collected through the wireless communication in order to estimate the amount of need at the base. In many cases, the surveillance reports cannot be sent directly to the base/sink and they require a multi-hop relay path. It is life-critical to send surveillance data without delay. The key issue is to avoid accessing a node called stuck node of the “local minimum phenomenon” [1] which causes detours and wastes time. A detour-free multi-hop routing, which is also called progressive routing, requires each hop to advance to a closer successor to the destination. The progress routing not only avoids any unnecessary detour delay, but also allows more concurrent reporting processes in the networks when fewer nodes are involved in the transmission. Note that a progressive routing does not necessarily have the shortest path due to the redundant neighbors available in node selection. In real environment, the occurrence of detour can be caused not only by “deployment holes” such as sparse deployment and physical obstacles, but also by many dynamic factors in real environment, including node failures, signal fading, communication jams, power exhaustion, interference, and node mobility. In order to achieve reliability and scalability in dynamics, the path in progressive routing is built by the independent decision of each intermediate node that selects the successor from its 1-hop neighbors. This selection relies on accurate information to predict all the candidates in its succeeding paths and then to know whether all them are available. Such capability information can guarantee each hop to advance along a progressive routing path. Our work provides each node this required capability information in a proactive manner with a structural regularity for all different paths passing through, saving the cost and delay of reconstituting the probing process in the reactive model (e.g., [9]). However, the neighborhood connections at each node are of irregular structure. A relay node will have different successor candidates, as well as their availability under the impact of local minima, every time when its relative location to the destination changes. Consider the availability status of node u3 in Fig. 1. In the routing from s to d, using node u1 will cause the routing to be blocked at node u4, which is called a stuck node. In this case, not only the stuck node u4, but also its nearby nodes u1, u3, and u4 must be excluded from the access of the routing because their succeeding paths are blocked. However, when the routing from u7 to u4 is initiated, the access of u3 must be allowed to keep the routing progressive. Those existing methods (e.g., [9, 11]) in the reactive model require to collect the information from the entire network in an ondemand manner to ensure the node capability. They face the problem of delay and cost in reconstituting the information for each newly initiated routing. Exiting proactive models (e.g., boundary model [6] and convex area model [2, 4, 5]) are not precise enough to catch such a dual role of node in each case. Even though many nodes become capable to successfully forward the packet in progressive routing, they will still be disabled from the consideration of routing decision as well as their communication ability.