View article

To correctly examine ad hoc network protocol performance, node mobility must be accurately modeled. Many ad hoc networking studies model node mobility using the random waypoint or random walk mobility models because they are easily simulated and might be assumed to represent a worst-case for node mobility. These models, however, do not represent a worst-case in terms of network performance. This paper presents a rigorous simulation of network performance in ad hoc networks using the random waypoint and Manhattan mobility models using the ns-2 network simulator. This work models the effects of buildings without restricting radio propagation distance, first by increasing block size to prohibit direct communication between nodes across blocks, and then by implementing a new radio propagation model in ns-2 that simulates the urban canyon effect of metropolitan areas. While there is little impact on network performance between mobility models without these enhancements, when modifications are made to simulate buildings and urban canyons, network performance suffers greatly and it is clear that the mobility model has a significant impact on simulated network performance. In addition, the results show that the impact of the choice of mobility model lies in the geographic constraints placed on node locations, not necessarily the nodes’ mobility.