Mobility Sensitive Broadcast Algorithms in Highly Mobile Ad Hoc Networks
Hao Zhang and Zhong-Ping Jiang
Broadcast is an important function in mobile ad hoc networks. Conventional flooding technique is known to be very expensive and results in unnecessary redundancy, high energy cost, packet collisions and con-tentions. More efficient broadcast schemes are proposed in the literature to alleviate this so called broadcast storm problem, and they have been proved to be able to improve network efficiency by reducing unnecessary retransmissions. However, most of these schemes do not consider the mobility factor, which can deteriorate network performance. We consider neighbor-designating-based broadcast schemes and self-pruning schemes, and propose improved algorithms for each category based on mobility analysis. Three mobility scenarios of nodal movement are attacked in this paper, i.e., the movement of one-hop neighbors, two-hop neighbors, and the entering of new nodes. We use a notion of relative stability level to describe the robustness of the links between a node and its neighbors, which can be predicted by location or neighbor information. For neighbor-designating- based broadcast schemes, instead of selecting those one-hop nodes that cover the largest number of two-hop neighbors, we choose those nodes that have more stable links with the sender. Furthermore, we use some threshold to ensure the two-hop nodes have been covered by either one stable path (which is composed of all stable links) or more than one unstable paths. Also, a node will be selected as a forward node if it has been selected as a forward node at the previous transmission, but is an unstable neighbor of the previous sender and a stable neighbor of the current sender. This is to improve the packet reception probability of this node’s neighbors. At last, if a receiver receives a packet from a node that is not recognized as its neighbor, it will transmit the received packet. For self-pruning schemes, a simpler but similar mobility sensitive algorithm.