Let’s be honest. Like most things in this age of technology, it really is amazing that any of this works. All the applications that utilize wireless frequencies are incredible. From commercial radio and television to Wi-Fi, cellular, and radar; from civilian to military – everything working in harmony and not interfering with one another.
Well… almost, anyway.
For the past fifteen plus years that I have been involved in the wireless community, I have seen and experienced many things. However, there is one thing that I have only read about others’ encounters with: radar interference.
For those who are not familiar, the 5GHz frequency structure is broken down into four primary channel blocks, or groups. They are:
- UNII-1 (channels 36 40 44 48)
- UNII-2 (channels 52 56 60 64)
- UNII-2 Extended (channels 100 104 108 112 116 120 124 128 132 140 144)
- UNII-3 (channels 149 153 157 161 165)
Each channel is 20MHz wide and non-overlapping (unlike 2.4GHz where channels are actually 5MHz wide but are used in a 20MHz fashion so that the only real non-overlapping channels are 1, 6, and 11). When using wireless N or AC protocols for greater bandwidth, multiple channels are used (think Ether-channel in the Cisco world). So, in the UNII-1 group when using the N or AC protocols, the devices would be using channels 36+40 (plus some other protocol enhancements) thus providing greater speeds and throughput values. These values for the N protocol would be presented as say 44+ or 44-, meaning the base channel is 44 but is extending to 48 (the +) or to 40 (the -). The AC protocol actually uses four channels (80MHz wide) and is notated with an E, so it would be seen as 36E.
Now then, the UNII-2 and UNII-2 Extended are special as together they make up the DFS group (Dynamic Frequency Selection). The DFS group is a portion of the 5GHz spectrum that is shared between free, unlicensed Wi-Fi, weather radar, and airport radar. Radar takes precedence, so in areas where radar is present, DFS cannot be used for Wi-Fi purposes.
This concludes your Wi-Fi mini lesson for the day. I hope you took notes because there will be a quiz later.
One of the deployments that I manage is an outdoor mesh network. This network connects various stations around a manufacturing plant’s wastewater treatment facility using 5GHz. The shortest link is about four hundred meters while the longest is about 1500 meters (or just short of a mile for us U.S. folks). About six or seven months ago, we had to replace the longest link radio. This radio had an internal semi-directional antenna but was replaced with a radio with an internal omnidirectional antenna as that was the only spare on site. The mesh link was weak but functional and carrying data.
Life goes on, as it tends to do whether we like it or not, and the link seemed to hold its own. In time, replacing the radio with another semi-directional no longer felt like a priority.
Several months later, it was determined that the replacement radio had not been added to the local monitoring system. Once that was corrected, we started to see random drops and connection instability. Sometimes there would be just one drop; sometimes there would be several. Sometimes these would happen back to back and others would be hours apart. The connection drops happened mostly at night, but on occasion would happen during the day as well and were almost always the replacement radio with the omnidirectional antenna.
This was some of the strangest activity I had ever seen. Something would cause a link drop, and then due to the replacement radio having a weak connection at best, there could be several hours of overall link instability.
After reviewing settings multiple times, watching specific links over time, and basically pulling my hair out trying to figure out what was happening, I decided to dig into the radio logs (which, admittedly, I should have done long before this) and there it was…
local1.warning,2020,10.X.X.X sapd: <404076> <WARN> <B4:5D:50:C9:3A:84> AM b4:5d:50:13:a8:50: Radar detected on interface wifi0, channel 116,typeid 5
local1.warning,2020,10.X.X.X sapd: <404088> <WARN> <B4:5D:50:C9:3A:84> AM b4:5d:50:13:a8:50: ARM HT Radar Detected Trigger Current Channel old 116+ new 140+/8
local1.warning,2020,10.X.X.X sapd: <404076> <WARN> <B4:5D:50:C9:3A:84> AM b4:5d:50:13:a8:50: Radar detected on interface wifi0, channel 124,typeid 7
local1.warning,2020,10.X.X.X sapd: <404088> <WARN> <B4:5D:50:C9:3A:84> AM b4:5d:50:13:a8:50: ARM HT Radar Detected Trigger Current Channel old 124+ new 153-/6
Right there in front of me in black and white, the logs told me exactly what was happening. I had never personally seen this before. I had only read about it in articles or on message boards.
I started pouring through all the logs I had for this group of mesh radios. I was able to correlate most of the big link drop and instability issues recorded by the monitoring system to various radar hits in the log files. When going back and looking at the longer disturbance periods, I would see the radar hit and the channel change, but the new channel would be another DFS channel. Moments later, there was another radar hit in the DFS channel the radio had just moved to and the drop and instability started all over again. It was surreal and it was all there in the logs.
Upon further investigation, many of the other mesh radios showed radar hits as well, but due to the strength of their connections, after a radar hit and subsequent channel shift, they never dropped their link nor was data communication significantly disrupted.
After this discovery, I was able to talk with the on-site team and explain everything that I had found. I told them that the fix was actually very easy, now that we knew what was happening. We needed to take the DFS channels out of the allowed channel selections in the radios’ management system. If they cannot use the DFS channels, they cannot be disrupted by radar.
The next day, I logged into the system and removed the DFS channels from the radios. The vast majority of drops and link instability disappeared. There was still a little due to the weakness of that long link, but a new radio with an internal semi-directional antenna has been ordered to replace the omnidirectional to restore the longest link to its former glory.
So, there it is – my first experience with radar interference.
If you have enjoyed what you have read, follow this blog as I share my experiences, blunders, how-to’s, tips, and opinions in all things OT Wireless from the wonderful world of industrial Wi-Fi!