An efficient hybrid model and dynamic performance analysis for multihop wireless networks

Multihop wireless networks can be subjected to nonstationary phenomena due to a dynamic network topology and time varying traffic. However, the simulation techniques used to study multihop wireless networks focus on the steady-state performance even though transient or nonstationary periods will often occur. Moreover, the majority of the simulators suffer from poor scalability. In this paper, we develop an efficient performance modeling technique for analyzing the time varying queueing behavior of multihop wireless networks. The one-hop packet transmission (service) time is assumed to be deterministic, which could be achieved by contention-free transmission, or approximated in sparse or lightly loaded multihop wireless networks. Our model is a hybrid of time varying adjacency matrix and fluid flow based differential equations, which represent dynamic topology changes and nonstationary network queues, respectively. Numerical experiments show that the hybrid fluid based model can provide reasonably accurate results much more efficiently than standard simulators. Also an example application of the modeling technique is given showing the nonstationary network performance as a function of node mobility, traffic load and wireless link quality.