Harry Jessell over at TVNEWSDAY has this story about a possible interference problem between operation of 3.65 GHz band equipment and the neighboring C-Band satellite receiver operated by CBS-owned KYW in Philadelphia.
According to the article, KYW experienced interference on its C-Band downlink near the 3.70 GHz frequency in February 2008, and resolved the problem by shifting to a higher frequency. The interference stopped a short time later, then flared up again in September, prompting KYW to call the FCC. That seems to have taken care of the problem, indicating it was a byproduct of some human operation addressed by the FCC enforcement — although possibly not. According to the article, the FCC won’t talk about it — which is standard procedure in an enforcement complaint.
According to the article, KYW Chief Engineer Rich Paleski thinks the problem was a “WiMax operator” using the 3.65 GHz. Paleski worries that 3.65 GHz will not be compatible with C-Band satellite downlink operation and warns “that should concern every station that imports programming via C-band satellite, which is to say just about every station in the United States.” He wants all television broadcast engineers to be alert for interference in the lower part of the C-Band near 3.70 GHz.
Given the rule limitations on use of the 3.65 GHz band, I am extremely skeptical of Paleski’s conclusion. Why? Because given the rules for operation in the band, no one should have been operating on the band in Philadelphia. And even they were operating illegally, they would have needed to hack the equipment to get within 25 MHz of 3.70 GHz, or have anything like the power needed to cause the kind of interference Paleski reports.
Given the growing popularity of the 3.65 GHz band for WiMax (as evidenced by projects like these), I think it’s important to look at this very carefully and not go leaping to conclusions. The 3.65 GHz band holds out a lot of hope for rural broadband by wireless ISPs (WISPS) running small businesses and priced out of licensed spectrum. Before anyone starts speculating from this single incident that use of 3.65 GHz poses a danger lets take a careful look at some of the facts around the use of 3.65 GHz and why I don’t think this is an industry-wide issue. It’s always easy to blame the new neighbor — especially when you think “their kind” is trouble. But how likely is it really?
More below . . .
Lets review the history of the band first. The FCC adopted final rules in the 3.65-3.70 GHz band (aka the 3650-3700 MHz band) in June 2007. You can read my summary at the time here, and a shorter, less personal version here. The FCC issued its first certification for equipment in the band in December 2007, pretty fast turn around time for a new band.
The timing between the first certified equipment in December 2007 and the first interference issue for KYW in February 2008 is tight, but possible. While it is unlikely that someone was using equipment certified for use in 3.65 GHz in February 2008, it is certainly possible. Mind you, it doesn’t explain what happened between February 2008 and September 2008, but nothing precludes it from being 3.65 GHz certified equipment so far.
The Exclusion Problem
The next problem we run into is that no 3.65 GHz system can operate in Philadelphia, or anywhere close. FCC Rule 47 CFR 90.1331 requires an exclusion zone of 150 km around this list of grandfathered satellite Earth-receiver stations. Philadelphia falls within the exclusion zone for Carteret, NJ, and is bounded in a ring of of exclusion zones in NJ, PA, WV, MD and VA. Since these zones are circles, it means that a legal 3.65 GHz system cannot legally operate within 300 km of Philadelphia. Since I am using approximations based on Google Maps, it looks like it is vaguely possible to have a legal system somewhere out near Cape May, NJ. So again, while we have not mathematically proven that a legal 3.65 GHz system, operating within FCC rules, could not have caused the interference to KYW in Phildelphia, it would take a pretty unlikely combination of circumstances.
But now we get to the last rule limitation which makes me even more skeptical that interference came from a legally operating system.
The 25 MHz “Guard Band”.
Paleski notes that the C-Band for satellite downlink abuts the top part of the 3650 MHz – 3700 MHz band. True. But, at the moment, only the 3650-3675 MHz portion of that can be used. This effectively creates a 25 MHz “guard band” between any existing device (especially one that existed in February 2008) and the bottom of the C-Band receiver range. Why? If you read my summary from 2007 you will note that one of the big issues on reconsideration was the between those favoring mesh networks and those favoring WiMax, or as I like to say, the MeshHeads v. the WiMax Posse.
Briefly, one of the big problems in the unrestricted bands is that you have many transmitters that operate all the time, while others use “contention based protocols.” They only operate when they have something to say. Before they transmit, they will determine whether anyone else is transmitting. If they detect activity on the band, they will wait a random amount of time before trying again. This is often called a “listen before talk” protocol. The standard IEEE 802.11 family of protocols, what we usually refer to as “wifi,” uses listen before talk. So do many mesh networks, since it enables multiple nodes to communicate easily without interference with one another and is a much more efficient way to use unlicensed spectrum. WiMax was developed for licensed services. It does not have a “listen before talk” protocol.
When the FCC considered whether to open the 3650 MHz band for non-exclusive use, one issue was how to make sure everyone who wanted to use the band could do so. In the pure unlicensed bands, such as 2.4 GHz band, we see an “arms race” where equipment manufacturers design equipment to generate signal all the time, shutting out the “listen before talk” devices. So the FCC required that any device certified to use the band must have some “contention based protocol” to ensure that everyone got to use the band and eliminate the incentive to have equipment that constantly broadcast for the purpose of blocking other equipment.
This made the WiMax Posse very upset, because WiMax has a very good reason not to have listen before talk — it was designed for a different environment. So they asked the FCC to eliminate the cntention based protocol requirement. The MeshHeads, including yre hmbl obdnt, told the FCC that the WiMax Posse already had plenty of space to play and if they didn’t like the 3.65 GHz rules they could stay in the other bands.
The FCC compromised. It agreed to pretend that WiMax was a “restricted” contention based protocol — meaning it worked when operators coordinated with each other rather than as a function of the technology (47 CFR 90.7). Devices using “unrestricted” contention based protocols (e.g., “listen before talk” type protocols) can use the entire band from 3650-3700 MHz. But WiMax and other “restricted” contention based protocols may only use the bottom half of the band, from 3650-3675 MHz (47 CFR 90.1319(c)). That creates a 25 MHz “guard band” between any WiMax operation and the C-Band operations at 3700 MHz.
Now here is the kicker. To the best of my knowledge (and certainly in February 2008), the FCC has only certified “restricted” contention based protocols. Why? Because WiMax is a very popular technology with a lot of pent up demand. But the IEEE did not approve a new unrestricted contention based protocol for the 3.65 GHz band — 802.11y — until September 28, 2008 — four days after the interference stopped. So while I expect equipment capable of using the upper half of the 3.65 GHz band soon, not a single device legally manufactured and sold in the U.S. for operation in the 3.65 GHz – 3.70 GHz band could operate on the 3.675-3.70 GHz half of the band when the interference happened.
This effective 25 MHz guard band also makes it extremely unlikely that it was a rogue system operating without a license inside the exclusion zone. The base stations are limited to 25 watts EIRP. To have come from a 3.65 GHz device, the person operating the device would have need to hack the device to either up the power limit or retune the device to function in the upper half of the band. Remember, to get an FCC certification to manufacture and sell a 3.65 GHz device usable for WiMax, you must demonstrate that the device physically will not work in the upper part of the band.
Why No Other Interference Problems?
Meanwhile, while there shouldn’t be a 3.65 GHz device operating within 300 km of Philadelphia, or within 25 MHz of the 3700 C-Band use, we do have a number of other places where we have seen deployment of 3.65 GHz equipment for licensed and legally operating systems. As far as I know,none of these systems has produced the same kind of interference with C-Band downlink receivers as experienced by KYW. As Peloski points out, virtually every broadcasters uses C-Band downlinks. Yet in all this time, not a single similar instance of C-Band interference has occurred where 3.65 GHz networks are up and running. But we are supposed to believe that a station in Philadelphia, surrounded by a 300 km exclusion zone, experienced interference in February and again in September from a 3.65 GHz transmitter with an out-of-band emission (OOBE) powerful enough to cause interference across a 25 MHz guard band?
I confess, I’m skeptical.
It’s always easy to blame the new neighbors. But before the broadcast industry and the satellite folks work themselves into a frenzy, they might want to consider some alternative explanations. True, the fact that the problem stopped (a second time) after calling in the FCC suggests some human agency (although if it starts up again in a month or two, maybe not). But it could easily be a malfunctioning piece of equipment creating nasty spurious emissions. Given all the restrictions on use of the 3.65 GHz band and the nature of the equipment, it seems more likely to me that the problem came from somewhere else than “those damn kids with their damn WiMax ruinin’ the neighborhood.”
Stay tuned . . . .