Impact of the Propagation Loss and mobility on the performance of AODV and DSR in MANETS

K. Amjad Radio Systems Research Group, Dept. of Engineering, University of Leicester, U.K. ka167@le.ac.uk

Mobile Adhoc Networks (MANETS) are formed by a collection of mobile autonomous nodes able to communicate with each other without any infrastructure. Node mobility and radio propagation dynamics make such networks challenging. Network simulation tools are commonly used to analyse the performance of MANETS protocols. The radio propagation models used for these simulations strongly influence the produced results. While the Two Ray Ground (TRG) model is the most widely used path loss model in reported MANETS performance analysis studies, it is not a realistic model for use in urban areas. In this study, we compare the performance of two widely used routing strategies for MANETS, i.e. Adhoc On-demand Distance Vector (AODV) and Dynamic Source Routing (DSR). Using the Network Simulator (ns-2), a variety of mobility models are incorporated in the same scenario (e.g. two group movements and three random mobility patterns) in order to mimic the typical high street mobility behaviour and several different radio propagation models were used to analyse the performance of an ad-hoc network with 100 nodes. Two new propagation loss models (i.e. Green-Obaidat Adhoc LoS model (GOA-LoS) [1] and ITULoS in street canyons (ITU-LoS) [2]) were incorporated into ns-2. The results from there were then compared with those produced using the TRG path loss model. The network performance is determined on the basis of packet delivery ratio (PDR), routing load and mean end-to-end delay with the effects of changing pause time (i.e. mobility level). The results indicate that the PDR is better if the communication channel behaves like TRG model for AODV and DSR protocols with varying pause time. The average PDR is about 7% & 10% better with TRG model than GOA-LoS and ITU-LoS models respectively. This is because of the extra losses incorporated in these models due to urban structures. With increasing pause time (i.e. lowering node mobility), DSR experiences higher routing load and mean delay if the channel acts like GOA or LoS-Urban model. . This is predominantly due to node density issues related with DSR as node movements, network congestion and propagation loss effects mainly invalidate the routes. AODV performs significantly better than DSR in all aspects irrespective of any propagation model. This study concludes that underestimating physical layer in MANETS may lead to more optimistic rather than realistic network performance. This study also verifies that node mobility and physical layer have a non-trivial impact on protocol performance in MANETS.

Figure1:AODVvsDSRPDRFigure2:AODVvsDSRRoutingLoadFigure3:AODVvsDSRMeanDelay

References: [1] D. B. Green, M. S. Obaidat, An Accurate Line of Sight Propagation Performance Model for Ad-Hoc 802.11 Wireless LAN (WLAN) Devices, Proceedings of IEEE ICC 2002, New York, April 2002. [2] ITU-R P.1411-5, Propagation data & Prediction methods for the planning of short range outdoor radiocommunication systems & radio local area networks in the frequency range of 300 MHz to 100GHz, December 2009.