We have a piecemeal wireless network made up of individual Netgears APs.
Today I did a survey of one of our buildings and was surprised to discover that was I was fairly consistently getting around 15% packet loss (with a 60 packet test) at most places in the building.
There was some channel conflict, essentially because there are 5 APs in this particular building.
Would channel conflicts be the likely cause of packet loss, or broken APs?
I believe it would drop packets if there is a conflict. What me and some work mates did when we had individual APs was speard them over the channel range and had them as far apart from APs the were on channls close to theirs. For example for you building with five AP in I would use channels 1,4,7,10,13 and 16 then have them placed around the building so 1 is at one end of the building and 16 was at the other.
I would also check the APs just incase one is broken. You may already do all this, but I hope it helps and is useful.
After fixing the overlap try looking for updated firmware and restarting them all. The hardware and software of the Netgear stuff can be quite squiffy. I'd be tempted to look into the avalibility of open firmware like tomato or open-WRT to replace the stock firmware with something a little more reliable.
I'm not a massive expert on wireless but I would disagree, I'd want channel 1 at one end and 4 at the other as these are very close (too close TBH) frequencies that may conflict.
Originally Posted by Grey-gear
Such high packet loss would have me wondering if you are bridging your switch side broadcasts over the wireless.
Fire up wireshark on a wired connection, and then on a wireless connection. If you see pretty similar packet rates and broadcasts etc then you will need to move your wireless APs (which are just acting as dumb-ish bridges) onto a separate VLAN with its own IP range, DHCP scope and default gateway.
On the topic of channel selection. The option put forth above regarding 1,4,7,10,13 and 16 is a disaster. ONLY use 1, 6, 11.
Why? Read this: Channels 1, 6 and 11 | MetaGeek and this: Adjacent Channel Congestion | MetaGeek
Note that the second linked blog post is actually half the explanation as to why Meru channel blankets work so well when intuitively they seem impossible. (the other half is to do with Meru's co-ordination of packet transmission between radios that can hear each other).