The new thing, not as good as the old thing

Recently we replaced the WiFi APs in some college accommodation blocks and immediately received complaints. The years old 802.11n units should be easily bettered by the brand new 802.11ac wave 2 APs so what gives?

We’re slowly but surely working through replacement our ageing estate of Aruba AP-125 units. The venerable AP-125 was a big seller for Aruba and, coupled with their 6000 series controllers, made for a high performing and attractively priced campus WiFi solution.

We had a lot of these APs, the model of choice when a number of new buildings went up on campus and whilst their quirky industrial design does look dated they’ve aged pretty well. Failures have been rare and although these APs don’t really have enough memory Aruba have kept them going through a lot of software releases by whittling away non-essential features like LLDP. However they die after software version 6.4.x.x and they’ve served us very well.

Anyway, from Aruba’s current range I selected the AP-314 with the AP-ANT-1W dipole antenna to replace the old dears because most of our AP-125s are wall mounted (the built in antennas can be rotated to suit the AP orientation). This makes for a larger and more imposing device but no matter, it’s got more antennas and spatial streams for MOaR WiFis!

The contractor installed the new APs in place of the old, moving them slightly where necessary to fit them in, and they were provisioned with the antenna gain specified of 6dBi for 5Ghz and 4dBi on 2.4.

On ArubaOS AP power output is specified as dBm EIRP. For APs with external antenna it’s necessary to give it calibration data by inputting the gain of the antenna for each frequency band otherwise a high gain antenna could lead to excessive and possibly illegal power output.

Immediately users began complaining that the upgraded WiFi was not as good as the old stuff.

Damn.

Fortunately a colleague and I had done a survey with Ekahau prior to the old APs being disconnected. We wanted to verify the existing AP locations were providing good coverage and found this was acceptable.

A return visit for a quick re-survey in a few rooms showed the coverage was no longer acceptable and backed up what the users were already telling us.

At this point I knew the AP locations hadn’t changed (at least in the first building I was looking at) and power levels were configured to be the same but coverage was definitely not as good.

So onto the bench (corner of a colleagues desk) with an AP-125, a 314 and for good measure a 134 using the same antenna.

A very simple setup using a Fluke Aircheck to measure the received signal strength with the AP being tested at a distance of about four metres. As far as possible the AP antennas were in the same place for each test, the aircheck left in the same location. APs were configured to use the same channel and power settings.

The results were pretty clear. Exactly the same signal strength from the two older APs with the AP-314 reading 4-5dB lower at 5GHz and 5-6dB lower at 2.4GHz.

I ran out of time to include an AP-315 in my tests, but it was clear there’s a problem with transmit on our newly installed APs.

Of course WiFi is a two way street so the question is whether this is a fundamental problem with the radio performance of the AP or, as I was beginning to suspect, an issue with transmit power.

I associated a client (iPhone 5) with an SSID on each AP and checked the SNR for the client. This showed the receive signal was roughly the same for all three APs, in fact slightly better on the 314.

A quick email to an Aruba contact revealed there’s been a change to how the transmit power is calculated based on the antenna gain. I’m not convinced this is the reason for the discrepancy I measured, as it would mean it isn’t possible to mix AP models within the same AP group… but whethr bug or feature, it needs fixing.

The short term answer has been to re-provision the AP-314s with much reduced antenna gain settings. This results in the AP-314 output matching that of the other AP models and hopefully keeping our users happy. Longer term it might be necessary to put all AP-314s in a separate AP group and configure higher transmit power – this would be very annoying.

As and when there’s any further information, I’ll update here.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.