Open Infrastructure
A Wireless Network Research Framework
for Residential Networks
Residential Broadband Network Study
OpenWiFi Testbed

We are deploying OpenWiFi routers in residential broadband networks to form a testbed for residential network study. Such testbed allows researchers to carry out a variety of experiments to profile the residential broadband network usage, learn the end users' Internet usage behavior, and to try out new research ideas in real residential networks.

The OpenWiFi routers are installed with our customized OpenWRT firmware.

  • The routers periodically report heartbeat message to our central mangement server, which includes a set of real time statistics data collected from each router.
  • Routinely, routers sniffer the users' network traffic and upload the data trace to our server for future analysis.
  • Besides, our central management server is able to remotely update the configurations of the routers, and schedule experiment tasks.
  • Currently, our testbed is running mainly in Boston urban area, spanning across 3 major ISP's.

    Data Traces:

    So far, we have collected 600GB residential users' traffic log. Our user samples consists of 15 residential broadband subscribers, including graduate students and young professionals.

    Data Trace Analysis Tools:

  • packet signature generator

    packet signature generator (pktsig) is an IP packet inspection tool designed for TCP/UDP data trace analysis.

    openwifi_pktsig accepts pcap data trace log(s) as input, and generates a "packet signature" for each IP packet. The packet signature captures as much information as possible to identify each IP packet, including the flow information, length, application type, etc. Given that HTTP traffic is now the predominant traffic in residential network, we also classifies the HTTP content type for each HTTP flow. openwifi_pktsig classifies the HTTP content type in a set of heuristics method, taking into account hostname, flow volume, etc., and obtains more accurate classification result compared with traditional MIME-type based classification.

    [Download]