Abstract: In this paper, we propose an accurate method to evaluate the available bandwidth in IEEE 802.11-based ad hoc networks. We improve the method described in. We evaluate our solution on di_erent scenarios. We also provide a comparison with different QoS routing protocols, BRuIT, AAC and QoS-AODV, based on di_erent available bandwidth estimation techniques.

---

Presented By:
Cheikh Sarr ”Claude Chaudet ” Guillaume Chelius ” Isabelle Guérin Lassous

---

Introduction
The quality of service issues in ad hoc networks are now extensively studied and more and more QoS protocols are proposed. Many protocols concern QoS routing. The goal of such a routing is either to provide the best routes in function of some parameters (like bandwidth, delay, packet loss, etc.) or to _nd routes that will o_er guarantees on some of these parameters. Many QoS routing protocols have been proposed so far and many of them consider the bandwidth parameter. To design an e_cient QoS routing, it is very important to get very accurate information on the used and available bandwidth. Such estimation is not so easy to compute in multihop wireless networks. The nodes share the medium and their perception of the used bandwidth or the available bandwidth can be very di_erent from one mobile to another. Therefore, before introducing a new _ow in the network, each mobile needs to precisely determine the available bandwidth that is o_ered to it, but it also needs to know the available bandwidth available to the nodes with which it may share the medium in order to not penalize them. Di_erent solutions, previously proposed, address this problem. They can be classi_ed into two main categories: the intrusive techniques that send probe packets for the estimation and the passive techniques that are based on a local computation of the available bandwidth and sometimes on a sparse exchange of this information. In this article, we claim that the proposed solutions are not satisfactory and that it is possible to improve the accuracy in the available bandwidth estimation for ad hoc networks. Indeed, the intrusive solutions consume too much bandwidth while the passive solutions mainly compute the available bandwidth with the formula _capacity minus used bandwidth_, which is not su_cient. Our solution is based on the IEEE 802.11 technology as it is widely used in wireless local networks and multihop wireless networks. However, our method could also be applied on other technologies as soon as they are based on a CSMA/CA approach, simply by using di_erent values for the parameters we use in our solution. We de_ne the available bandwidth between two neighbor nodes as the maximum throughput that can be transmitted between these two peers without disrupting any ongoing _ow in the network. This term should not be confused with the maximum throughput a _ow can achieve between two neighbor nodes. In we have proposed a method to evaluate the available bandwidth in ad hoc networks. If this method enhances the accuracy of the estimation, it is still unsatisfactory in some con_gurations. In this article, we improve this method. We also provide an evaluation of the technique by comparing it to several passive estimation solutions. Section 2 presents related work. Section 3 describes our available bandwidth estimation mechanism. Then, Section 4 evaluates and compares our solution with some other techniques.


read full report
http://hal.archives-ouvertes.fr/docs/00/...R-5935.pdf
http://caidaoutreach/isma/0312/abstracts/shah.pdf