Document Type


Publication Date



While the canonical behavior of today’s home Internet users involves several residents concurrently executing diverse Internet applications, the most common home configuration is a single external connection into a wireless access point (AP) that promises to provide concurrent highbandwidth Internet access for multiple clients through a wireless local area network (WLAN). Recent research has attempted to assess the performance impact of clients with weak wireless connectivity upon the other WLAN clients by employing measurement studies or analytic models that focus primarily on wireless channel characteristics. This paper examines the intertwined effects on performance of the user applications, the network protocol and the wireless channel characteristics via carefully designed measurement experiments that leverage previously developed network measurement tools. The study provides empirical evidence that suggests the overall performance of a wireless network is not only determined by the individual wireless channel qualities associated with each client, but also by the interaction of the various network layers with respect to transmission contention, queuing at the access point, the transport protocol, and the behavior of the specific applications. These results imply that effective WLAN performance modeling needs to include details on multiple network layers.