"Beamforming" (and the author is right to be skeptical of that term) is some really cool magic. My understanding of the history is that a lot of the technology comes from military radars. Development there was driven by power and compactness, and the relatively novel need (at the time) of building a radar that could both track targets and scan for new targets at the same time. Before that, you'd have a scan radar (antenna turning round and round), and a tracking radar (antenna pointed at the target). With beam forming tricks you can scan without moving the antenna, and then quickly jump back to get some additional tracking data. These days track-while-scan is easier because we can use computer models to do things like track interpolation, but back when this stuff was being developed that was much, much harder.
Twenty years ago when I was doing my PhD work, we were working primarily on passive radars (technically something called Passive Coherent Location (PCL)). With passive radars you have some transmitter of opportunity (a TV tower, or cell tower, or somebody else's radar), and want to observe the energy from that transmitter bounce off other things. One thing that's hard about that is that the transmitter is usually closer than the target, and the target doesn't reflect all its energy, so the returns from the target are many order of magnitude weaker than the direct return from the transmitter. So you need to design your antenna or antenna array very carefully to get only a small amount of the transmitter power (don't want to saturate your analog side), while getting as much of the target power as possible. You can do that with static antennas with "manual beamforming", but dynamic approaches allow you to correct for things like multi-path and changing atmospheric conditions.
Our particular work was in very low cost versions of this, with the end goal of spreading transmitters all around ("netted radar"). Good for cost, good for sensitivity, good for counter-stealth, but meant a very limited budget for any station. With the technology available now for building antenna arrays, we could have done so much better than we did back then. Amazing how the technology has changed in just a couple decades.
This seems to have been driven by consumer products. Wifi, cell phones, bluetooth, etc. The miniaturization, cost reduction, power reduction, etc. The technology came originally from big military radars, but what's driven the revolution seems to be the volumes and demands of consumer products. That's definitely going back into military and high-end civil technology, which is going to drive even more interesting requirements. A fun case study in technology.
(I have no non-public information on military radar technology, US or otherwise.)
I worked on applications similar to your PCL: Coherent backscatter systems for long-range RFID. It was really fun to apply "old school" techniques such as phased array direction of arrival and FMCW ranging for passive RFID tags.
What used to consume entire tabletops is now condensed into exceedingly capable software defined radios. It's insane what you can get for a couple-$k from places like Ettus: https://www.ettus.com/
Whoa! Went to the site, and out of curiosity open their "USP Product Selector" page. When I saw that it's just a Google Form (no instant feedback), and it starts at page 1 of 75, I bailed.
A monopulse tracking radar can track a target within hundredths of a degree, with no more than a handful of components. You can build a monopulse sonar with two opamps.
The rule of thumb that seems to be common among my peers is that monopulse offers an angle accuracy improvement of roughly 10 to 1 for objects with nominal SNR (13ish dB).
So a radar with a real beam width of 1deg employing monopulse would result in an accuracy of 0.1deg if the object can be reliably detected.
Interesting. I've been out of this field for 20 years, and the hundredths of a degree came from my fuzzy memory. Happy to defer to your more recent knowledge.
Reads like a good intro into how beamforming works. One detail I found dubious is this part about bigger antennas:
> To stretch our non-mathematical metaphor well beyond its breaking point, a "bigger sphere" will contain more signal as well as more noise, so just expanding the size of your region doesn't affect the SNR. That's why, and I'm very sorry about this, a bigger antenna actually doesn't improve your reception at all.
In my understanding that is incomplete because it ignores amp noise. When your antenna is larger, you get more signal (yes including more noise), so you need to amplify it less before the signal can be sampled. That means less amplification noise and the effective SNR is better. Thus antenna gain, not just antenna directionality, will improve reception.
Larger antennas help even though both signal and noise are amplified. The difference is that former is more amplified due to coherence: the antenna is shaped in a way to add the signal parts in the same phase. On the other hand, the noise is added together in a non-coherent way, with different phases, which makes the result smaller.
Think about it as the difference between summing two unit vectors in the same direction - with a resulting vector of magnitude 2 - and doing the same thing with two ortogonal unit vectors - the result will have a magnitude sqrt(2).
The article does later get into "changing the shape of the scoop", meaning a more directional antenna. And the designs for directional antennas are generally larger than omnidirectional ones.
So I'm curious as to why UHF and VHF require larger antennas. My parent's house was about 100 miles from the WTC and the antenna was massive. People in the city though used smaller sized ones.
UHF and VHF (that is, Ultra High Frequency and Very High Frequency) are the 3GHz-300MHz range and the 300MHz-30MHz range, respectively.
That translates to a 10cm-1m wavelength for UHF and 1m-10m for VHF. Minimal effective antenna size for a basic wire antenna is 1/4 the wavelength, so you end up with big antennas for low frequency / large wavelength.
Note that different organizations use the same terms for slightly different bands. In the IEEE nomenclature for radar, UHF is 300MHz to 1GHz. 3GHz is solidly within S-Band. And L-Band is in between them.
You brought back a fun memory. When my family moved upstate, about 125 miles north of the city, my dad was determined to watch the yankee games on channel 11, and my grandfather wanted to see his Mets.
He special ordered a massive antenna at Radio Shack. This thing disassembled was the full length of an old full size Chevy station wagon, I’m guessing 12 feet in the box and a bit longer assembled. Dad and grandpa mounted the thing to a pole that was also quite high attached to a device that would rotate the antenna.
When it was time for a game, you had to rotate the antenna to hit a specific gap in the mountains, which varied for different channels. Good times.
most of the reason bigger TV antennas were a thing was the "bigger is better" illusion. They make fancy stacks of elements and pt "+28 Dbi" numbers on the box but at any specific frequency those could be outperformed by a simple single cut wire. (they weren't utter horseshit: considered across the whole band they could be better for being bigger, but thats a technical and theoretical "win")
There are places where "more metal harvests more signal" in antenna design, but they're little dots on a broad map, not where you'd expect them to be. Herringbone antennas at WiFi freqs are fun.
People closer to the transmitter receive enough raw signal power that they can get away with a really insensitive antenna. Farther away, the antenna needs to be more optimal to pick up enough signal to be usable. It's not that the giant antenna is magically better, it's that the smaller antennas are mediocre.
Every frequency has a wavelength, as in a literal length of wave, which means a literal length of perfectly sized antenna. There's a formula. The right size for FM (100mhz) antenna is something like 4 or 5 feet, but for wifi (2.4ghz), it's the size of a water molecule. That is not an accident, as the FCC did not lease 2.4 out permanently since water absorbs it, hence we can just plop an unregulated router/AP in our house. This paragraph is probably vague and generalized, and possibly wrong-ish towards the end, but close enough to aid general understanding.
Yeah antennas need to resonate, not just absorb. A litre of water wouldn't be a very good antenna at 2.4, but it would warm up due to absorption if it is a trong signal. A vhf antenna might make a terrible antenna at 2.4 as well, but it would also warm up in a strong enough signal. Don't belive me? Shove one in a microwave (please don't really).
> PS. Do not ever approach a radar, it can boil you alive!
"Fun" related story. For a year or so in grad school I was in charge of the microwave and radar lab. The safety folks came for an annual inspection, and I toured them around the lab. Lots of "this can cook you from across the room", "this makes 50kV", "this seemingly normal 220V 50Hz outlet is actually 115V 400Hz". That kind of thing.
A week later we got the inspection report. It recommended we upgrade our requirement for closed shoes to include steel toe caps. No other safety measures or signage needed.
We experimented with decentralized networks during our university classes and one of the professors described this, basically you can form a phased array with loads of small senders in perfect timewise sync, sending slightly delayed. Its very energy intensive though. The sending and the high resolution clockworks. (Similar to gps recieving).
Funny sidenode: We found out that almost every building has a huge central antenna, called elevator ropes & shaft. Attach a wifi router to that and everyone in the building can overcrowd that one central wifi.
utterly unrelated (but fun) anecdote: Had a "phased array" CB antenna setup consisting of 3 antenna elements, carefully placed and spaced on a mobile ground plane, and very carefully cut and coiled lengths of Big Chonky cable too hook them all up to the radio. It worked well; did Chicago to Colorado once on legal 4watt power.
But then someone gave us a "1000w linear amplifier" to try. This turned out to be a bad idea. the third antenna element glowed red a few seconds after key up and then the harness joints de-soldered. Best we could figure the SWR changes induced by the antenna whips moving around weren't big enough, at 4w input power, to even notice. but when stood up to larger potential the tiny misalignment caused by the tips of the whips waving were enough to result in significant direct power loss.
One addition on receiver side amplification: it is done all the time. Each piece of equipment has a noise floor that can be higher than the one in the signal (think weak signals but with a good SNR). A low noise amplifier at the receiver can improve the signal in that case.
Every receiver (perhaps the exception is a spark-gap receiver) has an amplifier. The article is trying to say that you can't improve the S/N ratio of a signal with an amplifier. You can always make it worse though. The aim is to make it less worse.
Wifi APs always operate independently. Even it has the same ssid and same local network. Each ap still talk to client by themselves. The illusion you moves between AP is done by 'disconnect to one and connect to other'
>An ideal "isotropic" antenna (my favourite kind, although unfortunately it doesn't exist) picks up the signal equally in all directions, which means the region it "scoops" is spherical.
It would be interesting to see (and test!) a spherical antenna -- if one does/could exist...
Also -- what might be interesting might be getting two spherical antennas of the exact same size in resonance, over a distance between the pair which is some exact whole number multiple of their diameter...
Or across pairs of such spherical antennas... or pairs of pairs...
The whole WiFi ecosystem today seems like answers in a search of questions. What we really need is not speed, but connection quality, interference resistance and, maybe, in the third place - range. For speed we have copper or optic fiber.
Already exists, it’s called MU-MIMO in the 802.11 world.
Edit: Just to expand on this a bit… typically, the nulls of an antenna’s radiation pattern are sharper (in terms of dB per degree) than the peaks (and this is true for both phased arrays and traditional antennas).
In other words, you can shape the beam to put one receiver in a very deep null while only slightly reducing the RSS at the other receiver. Do that for two beams at once, and each receiver gets (almost) as much power (from the intended beam) as they would if the beam were steered directly at it, while getting very little power from the other beam.
Adding more elements to the TX side enables a greater number of and deeper nulls.
You'd need to be connected to their wifi hardware to do that, since you wouldn't get interference between your beams and theirs without having two coherent sources.
Twenty years ago when I was doing my PhD work, we were working primarily on passive radars (technically something called Passive Coherent Location (PCL)). With passive radars you have some transmitter of opportunity (a TV tower, or cell tower, or somebody else's radar), and want to observe the energy from that transmitter bounce off other things. One thing that's hard about that is that the transmitter is usually closer than the target, and the target doesn't reflect all its energy, so the returns from the target are many order of magnitude weaker than the direct return from the transmitter. So you need to design your antenna or antenna array very carefully to get only a small amount of the transmitter power (don't want to saturate your analog side), while getting as much of the target power as possible. You can do that with static antennas with "manual beamforming", but dynamic approaches allow you to correct for things like multi-path and changing atmospheric conditions.
Our particular work was in very low cost versions of this, with the end goal of spreading transmitters all around ("netted radar"). Good for cost, good for sensitivity, good for counter-stealth, but meant a very limited budget for any station. With the technology available now for building antenna arrays, we could have done so much better than we did back then. Amazing how the technology has changed in just a couple decades.
This seems to have been driven by consumer products. Wifi, cell phones, bluetooth, etc. The miniaturization, cost reduction, power reduction, etc. The technology came originally from big military radars, but what's driven the revolution seems to be the volumes and demands of consumer products. That's definitely going back into military and high-end civil technology, which is going to drive even more interesting requirements. A fun case study in technology.
(I have no non-public information on military radar technology, US or otherwise.)