A taxonomy and evaluation for developing 802.11-based wireless mesh network testbeds
International Journal of Communication Systems
| Hi-index | 0.00 |
Characterizing interference is critical to understanding the performance of a wireless network. Many protocol and algorithmic work fundamentally depend on such characterization. However, current research considers interference models that are either over-simplified or too abstract with unknown parameters limiting their use in practice. We address this issue in connection with WiFi networks (i.e., IEEE 802.11-based) due to their widespread use. We first develop a practical, measurement-based model to estimate the capacity of any given link in the presence of any given number of interfering links in an actual deployed 802.11 network, carrying any specified amount of offered load. For a network with N nodes, only O(N) measurement steps are needed to gather metrics for individual links that seed the model. We provide two solution approaches: one based on direct simulation (slow, but accurate) and the other based on analytical methods (faster, but approximate). We also show that as a by-product of our research we can create a highly accurate simulation model (e.g., using a packet level simulator such as ns2) of a real deployed network by seeding the simulator with measurement data. In an application of the above-mentioned capacity model, we address the issue of supporting voice-over-IP (VoIP) calls in a wireless mesh network. Specifically, we propose solutions for call admission control (CAC) and route selection for VoIP calls. Call admission decisions are made by using the capacity model to predict whether the capacity constraints at various nodes will be satisfied if a new call is admitted with a given route. We also develop a polynomial-time algorithm to search for feasible routes. In addition to studying feasibility, we study several routing metrics such as shortest feasible path and maximum residual feasible path. The above modeling approach requires active measurements. Also, it requires instrumentation access to network nodes. These could be impractical in many deployment scenarios. To address this issue, we develop an approach to estimate the interference between nodes and links in a live 802.11 network by passive monitoring of wireless traffic using a distributed set of sniffers. We model the 802.11 protocol as a Hidden Markov Model (HMM), and use a machine learning approach to learn the state transition probabilities in this model using the observed wireless traffic traces. This in turn helps us to deduce the interference relationships. We show the effectiveness of this approach via simulations and real experiments.