Performance Evaluation: Two Flood-Cancellation Methods of the Blocking Expanding Ring Searches on the AODV/WiFi MANET Environment

Abstract

Blocking Expanding Ring Search(BERS) is the modified scheme of TTL-sequence based ERS(TTL-ERS) to increase the energy efficiency by adding a little latency and decreasing the number of route request packet(RREQ)s which are flooded into network to query the path during the route discovery process for the reactive routing protocol of Mobile Ad-hoc Network (MANET). Until recently, several variants of BERS have been proposed, and all of them use ‘chase’ packet to stop or cancel the flooding of the RREQ for fulfilled route request. In this paper, we modelled these main variants of BERS, classifying them into Source node Initiated Chase Packet (SICP) type and Reply node Initiated Chase Packet (RICP) type, and analyzed the performance of each scheme installed in AODV protocol over WiFi MANET environment by the NS-3 simulator. The results of this paper show undiscovered aspect, different from the results of mathematical and analytical studies or the results of network simulation performed on the low node density network: the performance value of the BERS series deteriorates dramatically after node density of the network comes high. It means that broadcasting of chase packet as well as broadcasting of RREQ is a large burden on the dense node WiFi network. Notably, this phenomenon is more distinct in the SICP type model (BERS, BERS*, BERS+), and it even leads to greater overhead than TTL-ERS in case of very high node density network. On the other hand, the RICP type models (tBERS, tBERS*, tBERS+), which are improved to start flooding cancellation earlier than SICP, exhibit greater performance improvement than expected in case of higher node density. These results show that, in the BERS series models, flooding cancellation by the broadcasting packet can be another factor of the broadcasting storm, and that the earlier the flooding cancellation, the greater the performance improvement effect than theoretical prediction in the high node density WiFi MANET.