This paper presents a path compression protocol for on-demand ad hoc network routing protocols, which is called dynamic path shortening (DPS). In DPS, active route paths adapt dynamically to node mobility based on the "local" link quality estimation at each own node, without exchanging periodic control packets such as Hello messages. Each node monitors its own local link quality only when receiving packets and estimates whether to enter the "proximity" of the neighbor node to shorten active paths in a distributed manner. Simulation results of DPS in several scenarios of various node mobility and traffic flows reveal that adding DPS to DSR which is the conventional prominent on-demand ad hoc routing protocol significantly reduces the end-to-end packet latency up to 50-percent and also the number of routing packets up to 70-percent over the pure DSR, in heavy traffic cases. We also demonstrate the other simulation results obtained by using our two novel mobility models which generate more realistic node mobility than the standard random waypoint mobility model: Random Orientation Mobility and Random Escape Mobility models. Finally, simple performance experiments using DPS implementation on FreeBSD OS demonstrate that DPS shortens active routes in the order of milliseconds (about 5 ms).

In this paper, we present a novel forwarding scheme to enhance communication stability based on geographic routing in mobile ad hoc networks, which is called "Position-based Heuristic Forwarding" (PHF). For alternative solutions to traditional ad hoc routings, many geographic routing algorithms have been proposed. Most of the existing routings impose a certain restriction, planarity, on the graph structure of network for delivering messages to destination definitely. PHF achieves the guaranteed packet delivery over Unit Disk Graph, which is more widely employed graph model for the study of ad hoc networks. Accordingly, to eliminate the restriction of the routing algorithms enhances the probability to deliver messages successfully in networks with high nodes' mobility rate. In the simulation of PHF, we have evaluated the performance comparisons between PHF and its related work, Greedy Perimeter Stateless Routing (GPSR) and Dynamic Source Routing (DSR), which are the prominent geographic and conventional topology-based routing protocols, respectively. The results show that PHF provides higher packet delivery success rate indicating better communication stability and equal or less overhead than these work.

In this paper, we propose an efficient route discovery scheme for mobile ad hoc networks called Hop-Wise Limited broadcast (HoWL). Since nodes do not identify the location of other nodes, some of the routing protocols proposed for mobile ad hoc networks use network-wide broadcasts to discover a route. In contrast, HoWL limits the area of a route discovery by predicting the current location of the destination node using history of hop counts of previously used routes. We also introduce Characterized Environmental Indicators (CEI) which characterize environments for networks of mobile nodes. Specifically, environments can be characterized by three indicators: node density, average hop count of utilized routes, and frequency of link failure. We have implemented HoWL as an extension to DSR on GloMoSim network simulator. Quantitative and qualitative performance comparisons were evaluated between HoWL and its related work, Expanding Ring Search (RING) and LAR. The simulations show that HoWL performs best when low communication overhead is desired where up to 20% decrease over RING and three-fold decrease over LAR in the overhead were demonstrated. Thus, HoWL is most effective in overhead sensitive environments such as battery-limited sensor networks.