A study of distributed QoS adapter in large-scale wireless networks

TL;DR

This paper proposes a distributed algorithm for supporting QoS in large-scale wireless ad hoc networks, dynamically regulating contention windows to prioritize multimedia flows without resource reservation, validated by Markov models.

Contribution

It introduces a novel distributed QoS support algorithm that dynamically manages contention windows and prioritizes multimedia flows in ad hoc networks.

Findings

Guarantees QoS for real-time and best-effort flows

Prioritizes multimedia flows using dynamic contention window regulation

Validated algorithm correctness with Markov models

Abstract

Considering the comfortably establishing ad hoc networks, the use of this type of network is increasing day to day. On the other side, it is predicted that using multimedia applications will be more public in these network. As it is known, in contrary to best-effort flows, the transmission of multimedia flows in any network need support from QoS. However, the wireless ad hoc networks are severely affected by bandwidth, and establishing a QoS in these networks face problems. In this paper, we have proposed a thoroughly distributed algorithm to support the QoS in ad hoc networks. This algorithm guarantees the QoS of the real-time applications vis-a-vis each other and best-effort flows as well. The algorithm suggested in this paper dynamically regulates the Contention Window of the flows and serves the flows in terms of their requests QoS choosing the smallest CW in every node. This algorithm also uses the fixed and/or less stationary nodes for the transmission of real-time flows by increasing the QoS of the multimedia flows. This algorithm is preferred because it prioritizes the flows that are of the same class but have not obtained favorite QoS compared to other flows of the same class in addition to classifying the flows in the network and offering better services to the classes of higher priority. All this occur without the controlled packets forwarding and resource reserving and freeing method. We have proved the correctness of this algorithm using Markov's mathematical model.