Skip to content
Cannot retrieve contributors at this time
48 lines (41 sloc) 3.78 KB


AGILE System

In this demonstration, we aim at showcasing the advanced adaptation capabilities of the AGILE system on top of commercial 802.11ac compatible hardware. The developed system builds on the open-source ath10k driver for 802.11ac chipsets of Qualcomm to deliver an abstraction of open and reconfigurable PHY layer. Our development provides an API enabling real-time reconfiguration of key operational parameters, such as channel central frequency per available wireless interface at the access-point (AP) side and channel bandwidth per connected client. Moreover, the API enables controlled band steering and client roaming capabilities, enabling efficient transferring of clients between the available interfaces of dual-band APs or even among multiple in-range APs belonging to the same network.

Building on top of the developed API exposing an open and reconfigurable PHY layer, the AGILE system also brings intelligent adaptation algorithms. Taking advantage of the inherent spectrum sensing capabilities of COTS 802.11 equipment, AGILE implements an accurate mechanism for quantifying spectrum occupancy, as impacted jointly by spectrum sharing and interference. The impact of common interference sources, such as co-channel 802.11 transmissions, cross-technology emissions (MW ovens, LTE-U, IoT devices) and even 802.11 impairments, such as “hidden-terminals”, can be characterized through a unified approach with minimal overhead. The outcome of the spectrum occupancy evaluation process is used to identify under-utilized spectrum fragments and drive efficient spectrum adaptation decisions (channel frequency and bandwidth) per AP interface through the underlying reconfigurable PHY layer.

In an effort to drive network-wide efficient resource allocation, AGILE also introduces a central controller entity that collects statistics from all in-network APs. Local agents run on each AP, as background services over the OpenWRT embedded Linux, to continuously estimate the link quality and traffic demand per client and also monitor the available channel airtime per AP interface. Information collected locally is fed to the central controller, where intelligent adaptation decisions are applied. Based on the collected link quality information per client, local controllers running at the APs enable identification of low-SNR links, which can be efficiently mitigated through bandwidth reduction and/or steering of clients using 5 GHz links to the 2.4 GHz band. In addition, network-wide traffic load information is exploited to detect performance bottlenecks that are dynamically handled by the controller through efficient distribution of available wireless capacity, by performing transparent to the end-users band steering and client roaming. Through the developed mechanism the overall network is able to dynamically adapt to varying channel and traffic conditions.

Alt text

Further details can be found in the published paper:


The measurement corresponding to a period of one month 12/05/16 - 16/16/16 can be in the dropbox link ( In addition, In each row, measurements are organized by the DC values as measured per MHz along with a datetime timestamp. As soon as you import it to your local SQL server, you will be able to run SQL queries, for example to characterize utilization over time of day or to compare utilization between weekdays and weekends, as shown in Figure 6 of the paper (