Difference between revisions of "Network quality measurements"

From OpenScience
Jump to: navigation, search
(Description)
(report fields completed)
Line 32: Line 32:
 
|-
 
|-
 
|id
 
|id
|
+
|Unique ID per report
|
+
|1…27463
 
|
 
|
 
|-
 
|-
 
|epoch
 
|epoch
|
+
|Unix time of the point when the measurement cycle started (usually 15 minutes after the full hour). Since all tests within a cycle were run sequentially, this is usually not the time of the actual test. The time of the actual test has not been recorded.
|
+
|1293837301…1306876501
 
|
 
|
 
|-
 
|-
 
|timestamp
 
|timestamp
|
+
|Start time of measurement cycle in local time (see ''epoch'' and ''timezone'').
|
+
|"2011-01-01 00:15:01"…"2011-05-31 23:15:01"
 
|
 
|
 
|-
 
|-
 
|timezone
 
|timezone
|
+
|Timezone for ''timestamp''.
|
+
|"CET, "CEST"
 
|
 
|
 
|-
 
|-
 
|timeindex
 
|timeindex
|
+
|"Hour of year" starting at 2011-01-01 00:00 (local time). Used to connect report table to topology table.
|
+
|0…3623
 
|
 
|
 
|-
 
|-
 
|node_id
 
|node_id
|
+
|Network node that was tested (ID is alias for IP address of the node and therefore may not be unique per node for nodes with multiple IP addresses).
|
+
|"n1"…"n353"
 
|
 
|
 
|-
 
|-
 
|duplicate
 
|duplicate
|
+
|For some measurement cycles, a particular node was select twice for test. ''duplicate'' is "yes" for the second test of this node and "no" in all other cases.
|
+
|"no", "yes"
 
|
 
|
 
|-
 
|-
 
|validity
 
|validity
|
+
|"valid" for all reports were the initial "up" tests succeeded, all ping commands, traceroute, and wget were executed successfully, and wget finished downloading 1 MiB of data. "incomplete" otherwise.
|
+
|"valid", "incomplete"
 
|
 
|
 
|-
 
|-
 
|test_status
 
|test_status
|
+
|0 if the initial "up" test succeeded, 1 if the test failed. In the latter case, all further tests were skipped and a new node was selected for test.
|
+
|0, 1
 
|
 
|
 
|-
 
|-
|ping_status
+
|ping*_status
|
+
|0 if ping command was executed successfully. 1 if ping command returned an error. A packet loss of 100% is considered to be an error (see ''ping_loss'').
|
+
|-1, 0, 1
|
+
|-1 if ping test was not executed (see ''test_status'')
 
|-
 
|-
|ping512_status
 
|
 
|
 
|
 
|-
 
|ping1k5_status
 
|
 
|
 
|
 
 
|-
 
|-
 
|traceroute_status
 
|traceroute_status
|
+
|0 if traceroute command was executed successfully. 1 if traceroute command returned an error. The latter never happened in the current data set.
|
+
|-1, 0
|
+
|-1 if traceroute was not executed (see ''test_status'')
 
|-
 
|-
 
|wget_status
 
|wget_status
|
+
|0 if wget command was executed successfully. Other status codes according to wget man page: 4=Network failure, 6=Username/password authentication failure, 8=Server issued an error response. Codes 6 and 8 may be caused by nodes which were not compatible with our testing method.
|
+
|-1, 0, 4, 6, 8
|
+
|-1 if wget was not executed (see ''test_status'')
 
|-
 
|-
|ping_loss
+
|ping*_loss
|
+
|Number of lost packets (a total of 100 packets were sent for each test). Note: for cases with 100% packet loss ping reports an error. For these cases you will find ''ping_loss'' to be -1 and ''ping_status' to be 1.
|
+
|-1,0…99
|
+
|-1 if ping* was not executed or returned an error (see ''ping*_status'')
 
|-
 
|-
|ping_rttmin
+
|ping*_rttmin
|
+
|Minimum round trip time (in ms) found in all received replies.
|
+
|-1, 0…28783.935
|
+
|-1 if ping* was not executed or returned an error (see ''ping*_status'')
 
|-
 
|-
|ping_rttavg
+
|ping*_rttavg
|
+
|Average round trip time (in ms) of all received replies.
|
+
 
|
 
|
 +
|-1 if ping* was not executed or returned an error (see ''ping*_status'')
 
|-
 
|-
|ping_rttmax
+
|ping*_rttmax
|
+
|Maximum round trip time (in ms) found in all received replies.
|
+
 
|
 
|
 +
|-1 if ping* was not executed or returned an error (see ''ping*_status'')
 
|-
 
|-
|ping_rttmdev
+
|ping*_rttmdev
|
+
|Standard deviation of round trip time over all received replies.
|
+
|
+
|-
+
|ping512_loss
+
|
+
|
+
|
+
|-
+
|ping512_rttmin
+
|
+
|
+
|
+
|-
+
|ping512_rttavg
+
|
+
|
+
|
+
|-
+
|ping512_rttmax
+
|
+
|
+
|
+
|-
+
|ping512_rttmdev
+
|
+
|
+
|
+
|-
+
|ping1k5_loss
+
|
+
|
+
|
+
|-
+
|ping1k5_rttmin
+
|
+
|
+
|
+
|-
+
|ping1k5_rttavg
+
|
+
|
+
|
+
|-
+
|ping1k5_rttmax
+
|
+
|
+
|
+
|-
+
|ping1k5_rttmdev
+
|
+
|
+
 
|
 
|
 +
|-1 if ping* was not executed or returned an error (see ''ping*_status'')
 
|-
 
|-
 
|wget_transfer
 
|wget_transfer
|
+
|"aborted" if wget reported a "read error" during the download. "completed" if the download finished successfully.
|
+
|"aborted", "completed"
 
|
 
|
 
|-
 
|-
 
|wget_bytes
 
|wget_bytes
|
+
|Number of bytes transferred by wget until wget exited. The wget download may have stopped either because the download was complete or because it was interrupted (e.g. timed out). See ''wget_transfer''.
|
+
|-1, 40…1048576
|
+
|-1 if wget command was not executed successfully (i.e. ''wget_status'' is not 0)
 
|-
 
|-
 
|wget_rate
 
|wget_rate
|
+
|Download data rate as reported by wget (bytes per second). Note: wget reports only three significant digits ("180 B/s", "1.18 MB/s"), the number is converted to bytes per second herein.
|
+
|-1.00, 160.00…419430400.00
|
+
|-1 if wget command was not executed successfully (i.e. ''wget_status'' is not 0)
 
|-
 
|-
 
|traceroute_hops
 
|traceroute_hops
|
+
|Number of "hops" (on IP level) between measurement node and tested node.
|
+
|-1, 1…19
|
+
|-1 if traceroute information is missing or incomplete
 
|-
 
|-
 
|traceroute_pathlq
 
|traceroute_pathlq
|
+
|Product of all LQ values along the path reported by traceroute. This is the probability of a successful packet transmission from the tested node to the measurement node (from OLSR's point of view).
|
+
|-1, 0.000…1.000
|
+
|-1 if LQ product could not be computed (e.g. missing information)
 
|-
 
|-
 
|traceroute_pathnlq
 
|traceroute_pathnlq
|
+
|Product of all NLQ values along the path reported by traceroute. This is the probability of a successful packet transmission from the measurement node to the tested node (from OLSR's point of view).
|
+
|-1, 0.000…1.000
|
+
|-1 if NLQ product could not be computed (e.g. missing information)
 
|-
 
|-
 
|traceroute_etx
 
|traceroute_etx
|
+
|Computed ETX value for the path between tested node and measurement node.
|
+
|-1, 1.000…99.792, 999.000
|
+
|-1 if ETX could not be computed (e.g. missing information), 999 if either LQ or NLQ is 0 for at least one link ("infinity").
 
|}
 
|}
  

Revision as of 18:07, 12 September 2012

Description

Data from a performance measurement study carried out in a 802.11 wireless mesh network. The network uses the OLSR protocol in conjunction with the ETX metric. The study was conducted to evaluate the quality of service in terms of throughput, packet loss, latency, and jitter and the dependency of these parameters on the routing metric.

Contact to contributors

Datasets

Dataset Size Description
report.csv 27.4k lines, 1.1 MiB
traceroute.csv 101k lines, 346 KiB
topology.csv 3.1M lines, 19 MiB

report

Column Description Range Special values
id Unique ID per report 1…27463
epoch Unix time of the point when the measurement cycle started (usually 15 minutes after the full hour). Since all tests within a cycle were run sequentially, this is usually not the time of the actual test. The time of the actual test has not been recorded. 1293837301…1306876501
timestamp Start time of measurement cycle in local time (see epoch and timezone). "2011-01-01 00:15:01"…"2011-05-31 23:15:01"
timezone Timezone for timestamp. "CET, "CEST"
timeindex "Hour of year" starting at 2011-01-01 00:00 (local time). Used to connect report table to topology table. 0…3623
node_id Network node that was tested (ID is alias for IP address of the node and therefore may not be unique per node for nodes with multiple IP addresses). "n1"…"n353"
duplicate For some measurement cycles, a particular node was select twice for test. duplicate is "yes" for the second test of this node and "no" in all other cases. "no", "yes"
validity "valid" for all reports were the initial "up" tests succeeded, all ping commands, traceroute, and wget were executed successfully, and wget finished downloading 1 MiB of data. "incomplete" otherwise. "valid", "incomplete"
test_status 0 if the initial "up" test succeeded, 1 if the test failed. In the latter case, all further tests were skipped and a new node was selected for test. 0, 1
ping*_status 0 if ping command was executed successfully. 1 if ping command returned an error. A packet loss of 100% is considered to be an error (see ping_loss).
traceroute_status 0 if traceroute command was executed successfully. 1 if traceroute command returned an error. The latter never happened in the current data set.
wget_status 0 if wget command was executed successfully. Other status codes according to wget man page: 4=Network failure, 6=Username/password authentication failure, 8=Server issued an error response. Codes 6 and 8 may be caused by nodes which were not compatible with our testing method.
ping*_loss Number of lost packets (a total of 100 packets were sent for each test). Note: for cases with 100% packet loss ping reports an error. For these cases you will find ping_loss to be -1 and ping_status' to be 1.
ping*_rttmin Minimum round trip time (in ms) found in all received replies.
ping*_rttavg Average round trip time (in ms) of all received replies.
ping*_rttmax Maximum round trip time (in ms) found in all received replies.
ping*_rttmdev Standard deviation of round trip time over all received replies.
wget_transfer "aborted" if wget reported a "read error" during the download. "completed" if the download finished successfully. "aborted", "completed"
wget_bytes Number of bytes transferred by wget until wget exited. The wget download may have stopped either because the download was complete or because it was interrupted (e.g. timed out). See wget_transfer.
wget_rate Download data rate as reported by wget (bytes per second). Note: wget reports only three significant digits ("180 B/s", "1.18 MB/s"), the number is converted to bytes per second herein.
traceroute_hops Number of "hops" (on IP level) between measurement node and tested node.
traceroute_pathlq Product of all LQ values along the path reported by traceroute. This is the probability of a successful packet transmission from the tested node to the measurement node (from OLSR's point of view).
traceroute_pathnlq Product of all NLQ values along the path reported by traceroute. This is the probability of a successful packet transmission from the measurement node to the tested node (from OLSR's point of view).
traceroute_etx Computed ETX value for the path between tested node and measurement node.

traceroute

Column Description Range Special values
report_id
hop_id
from_node
to_node

topology

Column Description Range Special values
time
timeindex
this_node
other_node
lq
nlq

Publications

Currently in review.

Software

To be published.

Comments

None.