Performance Evaluation for WiFi DCF Networks from Theory to Testbed

Abstract

Distributed Coordination Function (DCF) is a basic MAC protocol used in the world-wide WiFi networks and plays a key role in determining the network performance, especially in situations with a large number of users and high-density Access Point (AP) deployed. To achieve a better understanding of the real-world performance of 802.11 DCF networks, we have constructed an emulation platform and a prototype testbed for performance evaluation. The design and implementation of these two platforms are discussed in this paper. The key DCF parameters, i.e., the initial contention window size ($CW_{min}$) and the maximum contention window size ($CW_{max}$), are tuneable so that we are able to study the impact of these DCF parameters on the network performance. These experiment results are compared against with a recently proposed unified analytical framework to examine the model assumptions and system performance bottlenecks. Our results demonstrate that by adapting the values of $CW_{min}$ based on WiFi traffic load, the maximal network throughput can be achieved, and the optimal value of $CW_{min}$ varies when the network size changes. As a reality check, the emerging software defined WiFi network architecture can be optimized for performance enhancement guided by this unified performance model.