Measurement-based Characterization of 802.11 in a Hotspot Setting

uw/sigcomm2004200610171020061017200508012006-11-14uw/sigcomm2004Dataset of wireless network measurement in SIGCOMM 2004 conference.

We are trying to understand how well 802.11 networks work in practice and how they can be improved. This dataset includes the traces collected by wireless monitoring and wired monitoring using tcpdump.

split a big pcap file into small (less than 1GB) pcap files and create a download directory for each traceuw/sigcomm2004/wireless2006-10-172004-08-302004-09-03293031323334http://www.cs.washington.edu/research/networking/wireless/http://www.crawdad.org/wiki/pmwiki.php?n=Main.Dataset.uw-sigcomm2004RFMON802.11 frames802.11802.11bpacket tracetcpdumpsignal strengthNetwork Performance Analysis802.11 infrastructureOur trace environment is the open wireless network provided to roughly 550 participants who attended the SIGCOMM conference in Portland, Oregon from 8/30/04 to 9/03/04. We view this as characteristic of a large and busy hotspot setting. The conference took place in a hotel with a layout depicted in the collection map below (in download url section).A wireless network comprised of 5 APs was set up for the conference, operating in 802.11b mode1 on channels 1, 8 and 11 (even though channels 8 and 11 are not orthogonal). Internet connectivity was provided by four separate DSL access lines. AP channels are as follows (see the collection map below (in download url section) for the location of each AP): channel 1: sigcomm-nat,sigcomm-public-3,sigcomm-public-1 channel 8: sigcomm-nat-foyer channel 11: sigcomm-public-2We monitored the activity on the wireless network in two ways. The first method was wireless monitors. The second monitoring method was traditional wired traces.MAC addresses and IP addresses have been anonymized using a tool from CRAWDAD project. Anonymization is consistent across trace files (IP-x maps to IP-y in all of them).1. One limitation that was beyond our control is that the SIGCOMM network was hampered by intermittent usability problems. We understand that these problems stemmed from the DHCP and DNS configurations and caused Internet connectivity to become unavailable to some clients. 2. A second limitation is that the different APs were assigned to logically different networks (with different SSIDs), such that users or their operating systems selected one AP for connectivity./download/uw/sigcomm2004/sigcomm2004-topology.png1520061017200508012006-11-14uw/sigcomm2004/wirelessTraceset of wireless monitoring.

Traceset of wireless monitoring.

split a big pcap file into small (less than 1GB) pcap files and create a download directory for each traceuw/sigcomm2004/wireless/chihuahuanuw/sigcomm2004/wireless/kalahariuw/sigcomm2004/wireless/mojaveuw/sigcomm2004/wireless/saharauw/sigcomm2004/wireless/sonoran2005-10-172004-08-302004-09-03Network Performance AnalysisThe first method was wireless monitors. We placed five PCs, each with three Netgear WAG311 wireless adapters, near the APs. Each monitor listened on all three active channels, using external antennas which we placed at least a foot apart from each other to avoid interference. Multiple monitors were within the range of each AP to provide information on the spatial diversity of transmission and reception. We logged all observed activity using tcpdump, capturing 802.11 control and management packets as well as data packets and writing complete packets to disk. The PHY information includes a receive signal strength indicator (RSSI) and the transmission rate. The MAC information is the entire 802.11 frame header and CRC. We also customized our MADWiFi driver to log reception errors that provide information on periods when transmissions were detected but not correctly decoded. Monitor channels (primarily) are as follows: ath0=1, ath1=8, ath2=11. Exception is that the APs switched from channel 8 to channel 6 on Wednesday morning (9/1/2004); kalahari switched ath1 to channel 6 Wednesday at 12:24 pm, and sahara, chihuahuan, and sonoran switched ath1 to channel 6 Friday at 8:30 am.1. We used the noise field of the prism frame to encode error status of frames, so be sure to filter down to only those frames with noise values of zero. others are errored packets (corrupt or not fully received). 2. The clocks on different machines were not synchronized; chihuahuan's clock is the most accuratethree of the ath2 traces (chihuahuan, mojave, and sonoran) were very large. the split command was used to break them up into 1G segments to allow sharing them online. you will need to cat the segments of each before unzipping.uw/sigcomm20044220061017200508012006-11-14uw/sigcomm2004/wireless/chihuahuanTrace of wireless sniffer (chihuahuan).

Trace of wireless sniffer (chihuahuan).

split a big pcap file into small (less than 1GB) pcap files and create a download directory for each tracefalse2005-10-172004-08-302004-09-03For hw/sw configuration, see the methodology section of the parent traceset uw/sigcomm2004/wireless. For the location of wireless sniffer, see the collection map linked in the dataset uw/sigcomm2004.tcpdump format (including Prism header). contents above the transport layer have been removed./download/uw/sigcomm2004/chihuahuanuw/sigcomm2004/wireless4320061017200508012006-11-14uw/sigcomm2004/wireless/kalahariTrace of wireless sniffer (kalahari).

Trace of wireless sniffer (kalahari).

split a big pcap file into small (less than 1GB) pcap files and create a download directory for each tracefalse2005-10-172004-08-302004-09-03For hw/sw configuration, see the methodology section of the parent traceset uw/sigcomm2004/wireless. For the location of wireless sniffer, see the collection map linked in the dataset uw/sigcomm2004.tcpdump format (including Prism header). contents above the transport layer have been removed./download/uw/sigcomm2004/kalahariuw/sigcomm2004/wireless4420061017200508012006-11-14uw/sigcomm2004/wireless/mojaveTrace of wireless sniffer (mojave).

Trace of wireless sniffer (mojave).

split a big pcap file into small (less than 1GB) pcap files and create a download directory for each tracefalse2006-10-172004-08-302004-09-03For hw/sw configuration, see the methodology section of the parent traceset uw/sigcomm2004/wireless. For the location of wireless sniffer, see the collection map linked in the dataset uw/sigcomm2004.tcpdump format (including Prism header). contents above the transport layer have been removed./download/uw/sigcomm2004/mojaveuw/sigcomm2004/wireless4520061017200508012006-11-14uw/sigcomm2004/wireless/saharaTrace of wireless sniffer (sahara).

Trace of wireless sniffer (sahara).

split a big pcap file into small (less than 1GB) pcap files and create a download directory for each tracefalse2005-08-012004-08-302004-09-03For hw/sw configuration, see the methodology section of the parent traceset uw/sigcomm2004/wireless. For the location of wireless sniffer, see the collection map linked in the dataset uw/sigcomm2004.tcpdump format (including Prism header). contents above the transport layer have been removed./download/uw/sigcomm2004/saharauw/sigcomm2004/wireless4620061017200508012006-11-14uw/sigcomm2004/wireless/sonoranTrace of wireless sniffer (sonoran).

Trace of wireless sniffer (sonoran).

split a big pcap file into small (less than 1GB) pcap files and create a download directory for each tracefalse2005-10-172004-08-302004-09-03For hw/sw configuration, see the methodology section of the parent traceset uw/sigcomm2004/wireless. For the location of wireless sniffer, see the collection map linked in the dataset uw/sigcomm2004.tcpdump format (including Prism header). contents above the transport layer have been removed./download/uw/sigcomm2004/sonoranuw/sigcomm2004/wireless16200508012006-11-14uw/sigcomm2004/tcpdumpTraceset of wired sniffer.

Traceset of wired sniffer.

the initial version2005-08-012004-08-302004-09-03Network Performance AnalysisThe second monitoring method was traditional wired traces, collected using tcpdump on the network segment connected to the APs. This provides a view of packets exchanged between the APs and the Internet.uw/sigcomm200447200508012006-11-14uw/sigcomm2004/tcpdump/greatvictoriaTrace of wired sniffer (greatvictoria).

Trace of wired sniffer (greatvictoria).

the initial versionfalse2005-08-012004-08-302004-09-03For hw/sw configuration, see the methodology section of the parent traceset uw/sigcomm2004/tcpdump. For the location of wireless sniffer, see the collection map linked in the dataset uw/sigcomm2004.tcpdump format (wired). contents above the transport layer have been removed./download/uw/sigcomm2004/greatvictoria-eth0.gzuw/sigcomm2004/tcpdump29uw/sigcomm2004tools/analyze/802.11/WitMaya Rodrigrodrig@cs.washington.eduUniversity of WashingtonDepartment of Computer Science and EngineeringGraduate student