IEICE TRANS. COMMUN., VOL.E93–B, NO.1 JANUARY 2010 29 PAPER A High Throughput On-Demand Routing Protocol for Multirate Ad Hoc Wireless Networks * Md. Mustafizur RAHMAN † , Nonmember, Choong Seon HONG † a) , Member, and Sungwon LEE † , Nonmember SUMMARY Routing in wireless ad hoc networks is a challenging issue because it dynamically controls the network topology and determines the network performance. Most of the available protocols are based on single- rate radio networks and they use hop-count as the routing metric. There have been some efforts for multirate radios as well that use transmission- time of a packet as the routing metric. However, neither the hop-count nor the transmission-time may be a sufficient criterion for discovering a high-throughput path in a multirate wireless ad hoc network. Hop-count based routing metrics usually select a low-rate bound path whereas the transmission-time based metrics may select a path with a comparatively large number of hops. The trade-off between transmission time and effec- tive transmission range of a data rate can be another key criterion for finding a high-throughput path in such environments. In this paper, we introduce a novel routing metric based on the efficiency of a data rate that balances the required time and covering distance by a transmission and results in in- creased throughput. Using the new metric, we propose an on-demand rout- ing protocol for multirate wireless environment, dubbed MR-AODV, to dis- cover high-throughput paths in the network. A key feature of MR-AODV is that it controls the data rate in transmitting both the data and control pack- ets. Rate control during the route discovery phase minimizes the route re- quest (RREQ) avalanche. We use simulations to evaluate the performance of the proposed MR-AODV protocol and results reveal significant improve- ments in end-to-end throughput and minimization of routing overhead. key words: high throughput routing, multirate routing, ad-hoc networks, AODV, on-demand 1. Introduction In recent years, the wireless ad hoc networks have become much popular for infrastructureless topology control and distributed collaboration among the wireless nodes. Each node in such a network has the ability to communicate di- rectly with any other in its communication range, while the out-of-range peers use intermediary hops to communicate with each other. The wireless ad hoc networks are applica- ble to a wide variety of fields as they are operable without any predefined infrastructure. A routing protocol in ad hoc networks dynamically controls the network topology, and hence, determines the network performance. As of now, many routing proto- cols have been proposed for ad hoc networks, where the on-demand protocols like DSR [1] and AODV [2] are Manuscript received February 23, 2009. Manuscript revised July 4, 2009. † The authors are with the Department of Computer Engineer- ing, Kyung Hee University, South Korea. ∗ This work was supported by the IT R&D program of MKE/KEIT (2009-S-014-01, On the development of Sensing based Emotive Service Mobile Handheld Devices). Choong Seon Hong is corresponding author. a) E-mail: cshong@khu.ac.kr DOI: 10.1587/transcom.E93.B.29 mostly preferred, because they save bandwidth and process- ing power [3]. In on-demand protocols, a node (source) searches for a new route to another node (destination), typi- cally by flooding a route request (RREQ) packet throughout the network. All neighbor nodes, excepting the destination, rebroadcast the received RREQ to its neighbors. When the destination receives an RREQ, it selects the route by send- ing a route reply (RREP) packet to the source in the reverse path. In the original AODV and DSR protocols, each node is expected to forward the RREQ only once (O(n) broad- casts). So, they forward the first received RREQ and discard the subsequent RREQ packets. Most of the available proto- cols are designed on single rate packet radio model and use hop-count as the routing metric. The expected transmission count (ETX) [4] and expected transmission time (ETT) [5] metrics are also proposed to select high throughput paths using link quality. Nowadays, the physical layer enhancements for wire- less communications support multiple data rates and enable nodes to select the appropriate transmission rate depending on the required QoS and the channel conditions. For exam- ple, the IEEE 802.11g PHY standard provides eight modu- lation and coding schemes (MCS) and thereby offers eight different data rates ranging from 6 Mbps to 54 Mbps [6]. A communication with high transmission rate requires a high signal-to-noise ratio (SNR) in the channel, and distance be- tween the sender and receiver is one of the key parameters that determines the SNR level, since the strength of a radio signal drops exponentially with distance. Thus, there is an inherent trade-off between high transmission rate and effec- tive transmission range [7]. MAC protocols, including the widely used IEEE 802.11 standard, do not include any spe- cific rate adaptation technique to utilize the multiple trans- mission rates efficiently, rather they leave it as an open issue for the vendors. Usually, a rate adaptation technique (eg. ARF [8], RBAR [9]) at the MAC layer selects the data rate for the next transmission from the recent success or failure history. However, the rate adaptation techniques cannot de- tect the RREQ failures (or, successes) because the broadcast and multicast packets are not acknowledged. For this rea- son, a node broadcast packets at a predefined data rate: usu- ally, at the highest or the lowest rate. Broadcasting RREQs at the highest rate covers the minimum area and increases the number of hops significantly. On the other hand, broad- casting at the lowest rate covers the maximum area, but it experiences excessive delay in the route discovery process and tends to select inefficient routes. Copyright c  2010 The Institute of Electronics, Information and Communication Engineers