Abstract

Airtime fairness, or time-based fairness, has been well recognized as a method to solve WiFi performance anomalies and provide a balance between fairness and spectrum efficiency in multi-rate wireless networks. However, the definition of airtime is vague and simplistic. In this paper, it is demonstrated that current airtime fair scheduling results in unfairness in reality because overheads are neglected or unfairly counted. We introduce a notion of responsible airtime, which covers not only the data transmission time, but also all overheads, even a TCP ACK segment in TCP traffic. An approach based on responsible airtime can provide true time-based fairness, but responsible airtime is too complicated to directly handle. A practical method is thus introduced for evaluating responsible airtime fairness indirectly via throughput measurement. The key element, throughput fair share, of a node, is based on the baseline property in time-based fairness. For each node, an achieving ratio of actual throughput to the throughput fair share is determined, and a new fairness index considering deficiency as well as equity is applied. To validate the feasibility of responsible airtime fairness, we have developed a simple responsible airtime fair scheduler in access points for download traffic. Extensive simulation experiments are conducted in various network and traffic environments using the ns3 simulator. The results show that true time-based fairness is achievable in practice.

Highlights

  • Since the data rate capability and channel condition of nodes may differ from each other in a wireless network, a WiFi network operates at multiple data rates

  • Most implementations of airtime fair scheduling are based on the deficit round-robin (DRR) [23], where the deficit is in the unit of bits, and the deficit counter of each node is reduced by its own quantum value, which is basically the weight of each node in weighted fair queuing [24]

  • Saturated TCP traffics were downloaded from the server in the wired side to each of the wireless nodes through the 802.11g access point (AP), where the wireless nodes were separated with a distance sufficient to break; // m_currentPacket refers a frame to transmit

Read more

Summary

Introduction

Since the data rate capability and channel condition of nodes may differ from each other in a wireless network, a WiFi network operates at multiple data rates. It is difficult to measure or analyze the actual responsible airtimes in a scheduling result, which are necessary to evaluate the time-based fairness of an implementation. In perfect time-based fairness, each node uses the same amount of responsible airtime regardless of the data rate. Most existing studies calculate airtime by analyzing its components, and fairness is implicitly ensured by allocating an equal amount of airtime to each node In this approach, the correctness of the fairness evaluation depends upon the precision and completeness of the analysis. We limit our development to a simple approximation of responsible airtime fair scheduling for download traffic It keeps measuring samples of time for which transmission of a data frame is completed, from requesting a medium access to receiving the corresponding MAC ACK.

Related Work
Necessity of a New Notion of Responsible Airtime
Analysis of Actual Airtime in Transmission
Typical
Responsible Airtime
Throughput Fair Share
Proposed Time-Based Fairness Index
Estimating the Responsible Airtime of a Frame
Responsible Airtime Scheduling
Validation of Fairness in TCP Traffic in a Basic Topology
Validation of Fairness in TCP
Fairness in Various Traffic and Network Environments
Conclusions
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call