Effect of Traffic Arrival Distributions on Routing Strategy in Multi-Hop Wireless Networks

Abstract

The basic problem of whether direct transmission or multi- hop routing increases goodput in multi-hop wireless net- works still lacks investigation from many aspects. This ar- ticle, approaches this problem by considering the effect of different traffic arrival distributions on the choice of rout- ing strategy for enhancing goodput of IEEE 802.11 DCF based multi-hop wireless networks under hidden terminal existence. Different traffic arrival distributions including Poisson, constant bit rate (CBR), Pareto and Exponential are considered, relaxing the generally adopted Poisson as- sumption, for various data rates over a wide range of traffic loads extending from unsaturated to saturated traffic loads. The goodput performance for all traffic arrival distribu- tions is found to be dependent on the traffic load in multi- hop networks. Of the four traffic models used, the network achieved the best goodput with Pareto and Exponential ar- rival distributions for light traffic loads, where CBR per- forms slightly better under heavy loads. The results suggest that a traffic load-aware pre-control mechanism on the traffic arrivals to the IEEE 802.11 MAC layer might provide signicant goodput gains in multi-hop wireless networks.