A Distributed Feedback Control Mechanism for Priority-based Flow-Rate Control to Support QoS Provisioning in Ad hoc Wireless Networks with Directional Antenna Dola Saha, Siuli Roy, Somprakash Bandyopadhyay Indian Institute of Management Calcutta Diamond Harbor Road, Joka Kolkata 700104 India {dola, siuli, somprakash}@iimcal.ac.in Tetsuro Ueda, Shinsuke Tanaka ATR Adaptive Communications Research Laboratories 2-2-2 Hikaridai, Seika-cho Soraku-gun, Kyoto 619-0288 JAPAN {teueda, shinsuke}@atr.co.jp Abstract— We have proposed a scheme for supporting priority- based QoS in mobile ad hoc networks by classifying the traffic flows in the network into different priority classes, and giving different treatment to the flow-rates belonging to different classes. We have adopted a control-theoretic approach to adaptively control the low-priority flows so as to maintain the high priority flow-rates at their desired level, thus guaranteeing QoS to high-priority flow. At the same time, our objective is to adaptively maximize low priority flows while maintaining high priority flows at a desired level so that full utilization of wireless medium can be achieved through adaptive rate control. To provide this desired service guarantee to high priority flows, we need a distributed flow-control algorithm. Here, the low priority flows, causing interference to a high priority flow, detect and measure high priority flow-rate at each node on their routes and consequently adjust their flow-rates using a feedback control mechanism to maintain the high priority flow at its desired level. This detection and measurement is done at MAC layer of each node participating in routing from source to destination. We have proposed this protocol with a very nominal overhead using omni- directional antenna and modified the scheme to show the overall improvement in throughput using directional antenna. The performance has been evaluated using QualNet network simulator and the results indicate the effectiveness of our scheme. Keywords-Ad hoc Network; Directional Antenna; Proportional Integral Derivative Control; Flow-Rate Control I. INTRODUCTION The recent progress in wireless communication and personal computing leads to the research of ad hoc wireless networks, which are envisioned as rapidly deployable, infrastructure-less networks with each node acting as a mobile router, equipped with a wireless transceiver. In this context, various solutions to QoS provisioning in mobile ad hoc networks have been proposed in recent past [1]. However, limited bandwidth of the mobile radio channel prevents giving every class of traffic the same QoS. So, some means for providing each class a different QoS must be implemented by assigning priority to one class over another class [2]. Though several solutions for wired environment are present, they do not work well in wireless ad hoc networks because of shared communication environment and host mobility. In this paper, we have proposed a scheme for supporting priority-based QoS in ad hoc networks by classifying the traffic flows in the network into different priority classes, and giving different treatment to the flow-rate belonging to different classes. We have adopted a control-theoretic approach to adaptively control the low-priority flows so as to maintain the high priority flow- rates at their desired level, thus guaranteeing QoS to high- priority flow. Several researchers have explored the idea of control theoretic approach for flow rate control in the context of wired network in order to control congestion in the network, to provide flow based end-to-end QoS as well as to deal with fairness issues. In [3], a control mechanism has been proposed that can be used to design a controller to support Available Bit Rate service, where users would dynamically share the available bandwidth in an equitable fashion, by adjusting an appropriate set of distributed controls based on feedback of explicit rates. Two flows in ad hoc wireless network will affect each other, when the two routes belonging to these two different flows share common nodes, or, they are close enough to interfere each other, causing route coupling [4]. In this case, nodes in those two routes will constantly contend for access to the medium they share. This is shown in Fig. 1. In such a situation, if the flow-rate of low-priority flow is reduced, the high-priority flow will get more chances to access the medium they share, which eventually reduces the congestion and improves the throughput of the high-priority flow. Thus, priority-based flow control is an effective means to provide service differentiation to different class of flows. Researchers have introduced end-to-end flow-control in transport layer to achieve service differentiation [5]. But, these schemes cannot ensure desired rates for high-priority traffic. Our objective is to adaptively maximize low priority flows while maintaining high priority flows at a desired level so that