Cisco Cisco Aironet 3500p Access Point Libro blanco
Copyright © 2010 Miercom
Cisco CleanAir Competitive
Page 16
the whole cell. A second test run never changed channel at 10 feet, took 2:10 to change at 50
feet, and 2:22 to change at 100 feet when a noise level of -70dBm triggered the threshold.
feet, and 2:22 to change at 100 feet when a noise level of -70dBm triggered the threshold.
The Aruba AP105 was also evaluated for its self-healing capability. The baseline noise level
read -105dBm. This reading was too low, and did not agree with the reading of the AP125 in the
same environment, which read -87dBm. In a network environment containing both AP105 and
AP125 devices, this mismatch in noise floor readings made it difficult to adjust the noise
threshold necessary to change channels. The noise level must be above the threshold for 120
seconds in order for a channel change to be triggered. After 30 minutes of lost clients due to the
video camera interference at the 10 feet location, the inaccuracy observed proved that the noise
level never remained above the threshold long enough to set the trigger. It was also observed
via the CLI interface that the AP kept resetting the radio.
read -105dBm. This reading was too low, and did not agree with the reading of the AP125 in the
same environment, which read -87dBm. In a network environment containing both AP105 and
AP125 devices, this mismatch in noise floor readings made it difficult to adjust the noise
threshold necessary to change channels. The noise level must be above the threshold for 120
seconds in order for a channel change to be triggered. After 30 minutes of lost clients due to the
video camera interference at the 10 feet location, the inaccuracy observed proved that the noise
level never remained above the threshold long enough to set the trigger. It was also observed
via the CLI interface that the AP kept resetting the radio.
In the 50 feet location, all clients lost ping when the camera was turned on. The AP105 reported
a noise level of -74 to -80dBm, but did not change channels over the 30 minute test duration.
a noise level of -74 to -80dBm, but did not change channels over the 30 minute test duration.
At 100 feet, all clients lost ping when the video camera was engaged. The noise reading on the
AP was -100dBm, and after 30 minutes of dropped clients, no channel change was observed.
We tried raising the setting for “Non 802.11 Interference Immunity” to Level 5 from the default of
Level 2, but all five clients remained unable to ping the access point.
AP was -100dBm, and after 30 minutes of dropped clients, no channel change was observed.
We tried raising the setting for “Non 802.11 Interference Immunity” to Level 5 from the default of
Level 2, but all five clients remained unable to ping the access point.
HP Equipment:
The smallest channel change interval for the HP access point is one hour. When the video
camera was turned on at 10 feet, the AP lost all clients. After more than an hour later, the AP
had not changed channel, nor logged anything in the event log. At 50 feet, only one client that
was closest to the AP remained on after the camera was turned on. Over one hour later, HP did
not change channel or log any events. At 100 feet, four clients remained connected, and only
the farthest client was jammed, as expected. Over one hour later, the AP had not changed
channel.
camera was turned on at 10 feet, the AP lost all clients. After more than an hour later, the AP
had not changed channel, nor logged anything in the event log. At 50 feet, only one client that
was closest to the AP remained on after the camera was turned on. Over one hour later, HP did
not change channel or log any events. At 100 feet, four clients remained connected, and only
the farthest client was jammed, as expected. Over one hour later, the AP had not changed
channel.
Trapeze Equipment:
Trapeze has a default scan interval of 3600 seconds, and the minimum scan time can be set to
900 seconds. With the video camera at 10 feet away, all clients dropped and Trapeze changed
channels after 47 minutes. At 50 feet, one client remained connected. After over an hour later,
Trapeze did not change channel. We noted that the noise level always reported -96dBm,
regardless of the position or distance of the jamming interference from the video camera. At 100
feet, only the farthest client was affected. After over an hour, Trapeze did not change channels.
900 seconds. With the video camera at 10 feet away, all clients dropped and Trapeze changed
channels after 47 minutes. At 50 feet, one client remained connected. After over an hour later,
Trapeze did not change channel. We noted that the noise level always reported -96dBm,
regardless of the position or distance of the jamming interference from the video camera. At 100
feet, only the farthest client was affected. After over an hour, Trapeze did not change channels.
Motorola Equipment:
Motorola AP-7131N offers legacy self-healing as well as the feature, Smart-RF. We enabled
Auto Channel Select and modified the data rate settings on the AP to increase available
bandwidth, and reduce channel utilization to support the video stream used in our testing.
Auto Channel Select and modified the data rate settings on the AP to increase available
bandwidth, and reduce channel utilization to support the video stream used in our testing.
With legacy self-healing, the AP uses the average number of retries as a trigger threshold to
change channels. With the video camera in the 10 feet location, Motorola reported 0 retries. It
could not detect any interference. The client throughput was low enough that it was represented
in scientific notation. After one half hour, the access point had not changed channels. The
Smart-RF feature was enabled and retested, but the same results were seen. No retries were
seen, and no noise level was reported. All statistics were zeroed out. The network was
completely jammed, but the access point could not detect it, and did not change channels.
change channels. With the video camera in the 10 feet location, Motorola reported 0 retries. It
could not detect any interference. The client throughput was low enough that it was represented
in scientific notation. After one half hour, the access point had not changed channels. The
Smart-RF feature was enabled and retested, but the same results were seen. No retries were
seen, and no noise level was reported. All statistics were zeroed out. The network was
completely jammed, but the access point could not detect it, and did not change channels.
At 50 feet, the average number of retries hovers between 1 and 2, and does not trigger the
threshold. After 20 minutes, the channel had not changed and we attempted to force ACS to
execute a channel change, but it did not occur.
threshold. After 20 minutes, the channel had not changed and we attempted to force ACS to
execute a channel change, but it did not occur.