1 Research Work
The so-called WiFi wireless networking technology was developed in the 1990s and at the turn of the century, was standardised under the IEEE 802.11 series of standards for Wireless Local Area Networks (WLANs). Beginning with physical layer bit-rates (PHY rates) of just 1 or 2 Mbps, over 20 MHz channels in the ISM band, IEEE 802.11 standards have evolved to giga-bits per second PHY rates in the recent IEEE 802.11ax and 802.11be standards. Thus, over the past two decades, the promise of WiFi technology being a wireless Ethernet appears to have been fulfilled. While this might be true for transmitter-receiver pairs, due to the way WiFi networks share the wireless bandwidth, there are inefficiencies and spatial reuse is unfair. WiFi equipment manufacturers aim to differentiate their products by overlaying various control algorithms (such as rate adaptation, dynamic channel selection, and dynamic aggregation, etc.) that attempt to provide better performance to users. Still the performance of various configurations are hard to predict, and can require substantial hand tuning to make the network perform well. In our larger project, the aim has been to develop a control strategy for performance management of arbitrary wireless networks. Towards such a goal, it is important to understand why WiFi networks perform the way they do. What are the reasons for the performance that various nodes in the network get when put together, although the performance is excellent when they communicate in isolated pairs? It is with this in mind that we have performed the research reported in this thesis. We take the approach of developing stochastic models for arbitrary WiFi networks. The models are based on an active set Markov chain, from which the required performance measures can be derived. Our primary purpose in this work is not to get a highly accurate estimate of the performance (which would require complex and intractable models), but to develop models that provide insight, correct qualitative behaviour, and reasonable estimates of performance. This work is divided into two part (i) Stochastic modeling of WiFi networks and (ii) Dynamic control of WiFi performance From ’Stochastic modeling of WiFi networks’ we will get a model for (i) Probability of collisions and (ii) Throughput of any complex Network. In the proposed model we are modelling practical IEEE 802.11 networks. This model can be used up to the IEEE 802.11g standards, and the IEEE 802.11n standard with MSDU aggregation. Moreover, this model can be used in both version of carrier-sense multiple access (CSMA) i.e, (i) CSMA Basic Version and (ii) CSMA/CA
2 Contribution to CNI
2.1 CNI workshop on WiFi QoS: Evolution of Standards and Performance Modeling, Feb 2021. I voluntarily participated in conducting a CNI workshop on WiFi QoS: Development of Standards and Performance Modeling in Feb 2021.
2.2 Networking Module Hands-on Laboratory Developed a Network module for the first year master’s students in the ECE department at IISc. Using this module effect of network critical parameters (like congestion window, packet lost probability, delay in the buffer, etc.) on throughput of the network are observe