There has been much prior work on WLAN channel assignment and power control. Several of them [9,24,14,10,7,21] either require modifications to the client or to the 802.11 standard. This makes them difficult to deploy. To the best of our knowledge, ours is the first proposal to be built and deployed that performs intelligent associations and deals with a dynamic operating wireless environment without requiring client modifications. Of the prior work in this area, we address two systems in particular that come closest to DenseAP.
Similar to DenseAP, MDG [10] identifies intelligent channel assignment, power control and client association as being key components of a systematic approach to increase the capacity of an 802.11 wireless network. It studies the interdependencies between these three knobs and identifies various situations in which a correct order of their application can increase network capacity. Furthermore, MDG modifies clients, and uses explicit feedback and cooperation from them to perform efficient channel assignment, power control and association. In contrast to MDG, DenseAP does not require any modifications to the clients, and therefore explores a different design space.
SMARTA [7] is similar to DenseAP in that it uses a centralized server to increase the capacity of a dense AP deployment without requiring client modifications. However, it uses a different approach. The central controller builds a conflict graph among the APs, and uses this graph to tune the AP's channel and transmit power. It does not manage client associations. There are two main differences between SMARTA and DenseAP. First, DenseAP relies on correctly managing client associations. We have shown that the benefits of a dense AP deployment is limited if clients are allowed to take association decisions. We have also shown that unilateral power control (without client cooperation) can hurt the performance of the system. We also note that since SMARTA is evaluated entirely in simulations, we are unable to do a fair comparison of SMARTA with our scheme.
In [6], the authors propose using a centralized scheduling mechanism to schedule downlink traffic in a dense deployment of APs. The overall goal is to efficiently manage the data plane of an 802.11 deployment. The work is in progress and at the time of this submission, the authors have not proposed a solution for managing the uplink traffic.
A host of products by networking startup companies [3,5,2,4,1] are designed to manage AP deployments in the enterprise. The exact details about how their products work are difficult to obtain. However, most systems seem to either ignore association control and load balancing, or they address such challenges by requiring users to install custom drivers.
NSDI-2008
