Search

1To mitigate performance anomalies due to different transmission rates and to improve throughput performance in ad hoc networks, previous work has proposed a new method called user-cooperative mobility. This article also shows that the method achieves good improvement in ad hoc nodes, i.e., human users appropriately move. However, this method only employed a theoretical evaluation, which does not include detailed behaviors such as the capture effect or dynamic back-off time, which are considered to largely affect the performance in terms of throughput. Here, we employ an NS3 network simulator to clarify how the difference in throughput affects user mobility and throughput improvement in an ad hoc network. The result shows a huge difference in system throughput between theoretical estimations and simulations. This is caused by different transmitting opportunities due to the capture effect and the dynamic back-off time. In some cases, the system throughput will even drop when the throughput improvement ratio is under 1.0, being 0.64 when the throughput results from user moves are compared with those from no moves. We conclude that to improve the throughput as much as possible, a user should move based on the simulation results, which includes a capture effect and back-off time, instead of using only theoretical evaluation.

To mitigate performance anomalies due to different transmission rates and improve throughput performance in ad-hoc networks, we propose a method called user cooperative mobility in which a node, i.e., a user of an ad-hoc network, moves (changes its geometrical position) for communication. Unlike previous research regarding user cooperative mobility, it is not easy to decide to where the user should move to obtain the best result because the move may change multiple transmission rates in multiple links. Thus, we show that the best position to move to is quite different from the intuitively best position on the way to the next hop. We first evaluate throughput characteristics at every position within a movable distance from the original position of the cooperative mobility user. Then, we compare differences in throughput and direction of movement between the following three movement strategies: the best move, the intuitive move and no move. In terms of throughput improvement, which is normalized by the throughput of no move, the best move results were 40% better than those of no move and 24% better than those of the intuitive move. The direction of movement significantly differed from the intuitive move, by 92 degrees at the maximum. We conclude that cooperative user mobility with the best move is beneficial for both cooperative mobility users and ad-hoc networks.