Please note: This master’s thesis presentation will be given online.
Midul Jacob, Master’s candidate
David R. Cheriton School of Computer Science
In this thesis we analyze data that was collected over a 24 hour period from 446 access points that provide connections for 2,946 clients. The data was obtained from deployments of modern commercial Google Wifi access points. A focus of this thesis is an analysis of the Received Signal Strength Indication (RSSI) from messages sent from clients to the access point. The RSSI depends on both the environment in which the signals operate and the distance between the client and the access point.
Our dataset includes 417,122 data points of which 45.1% of the data points are from signals using the 2.4 GHz spectrum and the remaining 4.9% are from the 5 GHz spectrum. The data has been collected by each access point (AP) every 5 minutes over a period of 24 hours. We find from our analysis that across all access points, the average number of clients that are simultaneously connected in any 5 minute window is quite small. That is, 65.7% of the APs have on average 3 or fewer clients that are simultaneously connected in any 5 minute window. However, we also find that 6.5% of the APs service on average 9 or more clients.
In this thesis we develop and utilize a methodology to categorize clients and networks using RSSI values (signal strengths) of the messages received by access points from the clients, to study the possible PHY rates which can be used by clients to send messages to the APs. The methodology also helps us to capture and examine the variability in signal strengths. Several previous studies have characterized WiFi networks using the measured throughput of the clients. However, the throughput experienced and rates used by clients in those studies depend on the capabilities of the clients. We believe that a significant advantage of our methodology is that it is independent of the capabilities of the clients used in the study. In addition, our methodology is also able characterize the environments in which the WiFi devices operate. This is because our methodology primarily uses the signal strength of the messages to characterize devices in a WiFi network and the signal strength changes over time due to the movements of the sender, receiver, or people in the area.
We use our methodology to analyze both clients and networks. We find from our analysis of clients that, over the 24 hour period, 90% of the signals from 84.2% of the clients are received by the APs with either Good or Moderate signal strengths. Thus, for the majority of the clients signal strengths are mostly quite reliable. We also find that clients using the 2.4 GHz spectrum have signals about as good as or better than clients using the 5 GHz spectrum. However, perhaps one of the most interesting findings is that, when analyzing networks we find that 27% or more of all networks have one or more clients whose signals are received by the AP with unreliable signal strengths. These clients could potentially impede the throughput of all the other clients in the same network and also networks in the vicinity, due to the WiFi performance anomaly problem. We also find that networks with more clients are not more likely to have clients with unreliable signals than expected based on probability. From the results of our analysis of clients and the analysis of networks, we note that a small number of clients may impact the performance of a considerably large number of networks.
To join this virtual presentation on Zoom, please go to https://us02web.zoom.us/j/86437950510?pwd=QTlZZGswM1d1dzhHdkRLUTQ3NWZEZz09.
Meeting ID: 864 3795 0510
200 University Avenue West
Waterloo, ON N2L 3G1