Recently I have been writing about the challenges I have experienced with industrial wireless compared to those discussed in the most recent issue IEEE Communications Magazine. Last week I wrote about multipath in industrial settings. This posts discusses interference in industrial equipment, which is something that certainly exists but has rarely posed a problem in installations I’ve seen.
The IEEE article says interference from industrial equipment presents a problem for industrial wireless. I never observed this problem personally. I would expect this problem to present as lost link budget, e.g. a link that should work with -70dBm signal strength requiring -50dBm.
If interference is intermittent, it may go unnoticed. Suppose the interference has 50% duty cycle and a period of more than few milliseconds. This halves the throughput. Usually industrial networks, especially if they’re industrial Wi-Fi, use a tiny fraction of the maximum throughput. So a channel being unusable half the time would not be noticed. Even if the channel is unusable for long periods of time, seconds to minutes, industrial users do not notice a problem unless something times out at a higher level and the HMI alerts operators.
A wideband signal is more susceptible to strong narrowband interferers. A narrowband interferer may take out only subcarrier of a 20MHz OFDM signal. The receiver can recover the data with a subcarrier missing, but not if it’s strong enough to saturate the ADC. The data is there, in this case, but the ADC doesn’t have the dynamic range to detect the weak OFDM signals in the presence of a strong interferer. At maximum data rate, the ADC in a Wi-Fi chipset has just enough dynamic range to digitize a Wi-Fi signal. There’s no extra dynamic range built in to deal with interferers that may take out only a few subcarriers.
Narrowband interference is less an issue for a narrowband hopper. When it hops to a frequency with interference, that transmission is lost, but data gets through on the next hop. They generally have less throughput, though, so there is less “headroom” before a reduction in throughput will be noticed.
When it comes to industrial Wi-Fi, most of my experience has been with 20MHz channels. I am very interested to hear how the 40MHz and 80MHz channels allowed by the new standards hold up in industrial settings. Except in isolated areas, I suspect effects of interference will offset most gains in throughput.
Latency
The biggest impact of interference is on latency. The radio must wait until the channel is clear. If a packet is not ACKed, it waits a random period before retrying. In industrial systems that require low latency, this is a huge problem. (Note: Some hoppers just blast out packets multiple times with no clear channel assessment and no ACKs.)
The biggest cause of latency issues is not interference but rather the channel sharing function itself. Next week I will discuss the issue of channel sharing in low-latency applications. This is the biggest issue of all for industrial wireless.