Earlier this week, an article titled Here’s What Wi-Fi Would Look Like If We Could See It got Slashdotted. This got my attention because, as a radio enthusiast that has been playing around with Wi-Fi “before it was cool”, I always had a mental idea of what Wi-Fi would look like. Unfortunately, the article itself is really just a bunch of pretty pictures and doesn’t really address the subject in any serious way. It provides a static visualization – one which doesn’t really mesh with reality for several different reasons (which I shall examine later).
First of all, I’m not a certified wireless professional – heck, I don’t even have a CCNA (at this time), but I am a radio amateur and I’ve spent some time looking at training materials for CWNA so I think I can try to do a better job at it.
The problem is that I’m not really a digital artist at all. I’m not skilled at computer graphics – so instead of actually providing you a visualization, I’ll provide you the ideas and you supply the imagination to make it work.
The first thing to address is the Wi-Fi frequencies and bands. At the moment, the most popular band is the 2.4Ghz band which spans from around 2401Mhz to 2495Mhz (depending on country – with 22Mhz width modulations, 11Mhz is below the centre frequency and 11Mhz is above the centre frequency). From this, we can see the entire 2.4Ghz band fits into 94Mhz, and each channel is about 22Mhz (for single width modulations) and up to 40Mhz for wireless N “bonded channel” type modulations.
The other band which is gaining in popularity at the moment is the 5Ghz band. Many countries have limitations in this band as to which range of frequencies are allowed, but the band itself spans from about 4904Mhz to 5836Mhz with a few gaps for weather radars. This leaves a total band width of 932Mhz.
A wavelength at 2.4Ghz is about 12.5cm and at 5Ghz is about 6cm. This is why the article says “Wifi waves are about 3 to 5 inches from crest to crest” and “A freeze frame of these pulses would show that the pulses are about 6 inches apart …” I suppose if you could freeze time enough to examine the individual waves, then you start to get strange results. But the fact of the matter is that we don’t perceive the individual waves of light themselves. Instead, we respond to the intensity (brightness) and wavelength (colour).
We already know that the visible spectrum extends from about 380nm to 750nm (or around 400 to 668Thz). Green light at 550nm is about 545Thz! A direct comparison of megahertz vs terahertz gives you an idea that a few megahertz is nothing, as a terahertz is one-million megahertz! So if nothing “scaled”, and our green receptors were instead sensitive to 2.4Ghz, all the channels would differ so little in colour that it would all appear monochromatic, kind of like the yellow low-pressure sodium vapor street lamps.
Quick recap of how our vision works – basically we have light-receptive cells called rods and cones. The graph shows a sensitivity peak at 420nm, 498nm, 534nm and 564nm. The curves actually overlap – but if we look at just the green and red ones, we can see that the peaks are about 30nm apart. The frequency is about 561Thz and 532Thz respectively. So we can easily sense a 5% frequency change.
So lets take this 5% back to 2.4Ghz – this is about 120Mhz. So there is hope. If we managed to have receptors which are peaking and overlapping much like our green-orange receptors, we would be able to perceive the 2.4Ghz band as a mix of colours ranging from green to almost orange depending on which channels are in use. Needless to say, as the 5Ghz band is almost 10 times greater in bandwidth, we should be able to resolve all of it if our cells were sensitive to this region.
Using the same analogy – the difference between the peaks of 420nm and 564nm are a difference between 714Thz and 532Thz which is a 34% change of frequency. 34% of 5Ghz is about 1710Mhz – so the 5Ghz band would be perceived as stretching from blue to about green.I know this is a bit of a fudge, but it’s an idea worth exploring for the sake of exploring it.
Anatomically, to be sensitive to 2.4Ghz or 5Ghz, it would be very difficult as your cells would need antennas about a quarter of the wavelength to be an efficient receiver …
Wi-Fi Power and Wave Behaviour
It’s no secret that Wi-Fi is a low powered RF signal. In fact, most of the wireless cards in laptops and wireless routers at home can emit no more than about 100mW. Putting this into perspective, an LED torch can easily be 2W (20 times stronger).What would 100mW of light look like? Well, that would depend on how it is spread.
Transmissions from dipole antennas (the basic stick antennas you get with your wireless card) spread light in a sort of doughnut like shape. You can think of this like a camping lantern – it’s 360 degrees. But 100mW is only the output power of around one to two 5mm LEDs. That’s … about one or two of the cheapest $2 solar garden lights.Of course, with higher gain Yagi antennas, you can focus them into tighter beams – most Yagis will give you about 15-25 degree beamwidths, so it’s like looking into a cheap torch with a single LED instead. A panel antenna would be like a solar garden floodlamp with a single cheap LED. There is a key difference though – unlike the reflectors within torches, the shaping of the beam isn’t exactly perfect, so there is some leakage out of the back or around the sides – known as sidelobes. These can take repeating patterns – you can see it on a polar plot of the antenna’s gain.
The interesting thing is that our eyes don’t allow us to see all the way around. So comparing our eyes to a dipole antenna is a bit of a problem – instead, our eyes are more like many Yagi antennas packed together. This gives us the ability to spatially resolve where the signals are coming from and form an image rather than just sensing “how bright the light is” (imagine replacing your eyes with a single phototransistor). So the way we perceive light is different to how our Wi-Fi cards would perceive the radio energy, although this is somewhat changing through the use of “beamforming” where the cards gain some ability to resolve signals through space (i.e. MIMO technology).
The 2.4Ghz and 5Ghz spectrum, unlike the visible spectrum, is quite quiet normally. There isn’t much that naturally contributes noise to the spectrum. This means that it’s normally quite dark. This means that even small amounts of light can be visible – just like your solar garden lights. The distance which the lights will be visible is comparable to the range of your wireless connection. A Wi-Fi card can easily handle anywhere from about -20dBm (antennas almost touching) to -94dBm (link just about to be dropped). This is a span of about 74dB. The human eye has a dynamic contrast ratio of about 1,000,000:1 which is 60dB. The human eye adjusts to the brightness by adjusting its pupil and chemicals in the eye – the wireless card adjusts to the strength of the signal through controlling its front end amplifier (by automatic gain control). So in most cases, the “light” can be seen about 30 meters at the most.
Lets take it one step further though – when it’s supposedly dark, it isn’t really. There’s actually a bit of background noise that always exists – this is the noise floor. When there is a lot of Wi-Fi transmitters around, you can expect the floor to increase – this is kind of similar to light pollution at night spoiling your view of the stars. It obscures dimmer objects and is like a fog or a haze.
The thing that seems to be quite neglected by the visualizations is the propagation of the radio waves that make Wi-Fi work. They’re not quite like light as we know it, although it is similar.
Radio waves are affected by many propagation phenomenon. They can be absorbed (like a black piece of paper absorbs visible light) and reflected (like a mirror or a white piece of paper reflects visible light). They can also be refracted across sharp edges (rarely visible unless you’re in a physics lab looking at interference patterns) and can add or cancel out (much like two outward rippling waves at a pond).
The key thing to recongize is that light travelling through empty space is ultimately not visible (in general). Just like if you shine a torch into the night sky – you can’t see the beam itself (and whatever you can is due to scattering). So the waves coming out of a router would actually be invisible even if you had Wi-Fi vision – save for some that gets scattered by molecules in the air, reflected by objects or if you were directly staring at the antenna. That’s one flaw in the visualization – you won’t see it until it “hits something”.
Another thing to realize is that objects which are opaque visually may not be opaque at microwave frequencies. Think about your bedroom wall – you can’t see through it, but the Wi-Fi travels through it just fine, with some loss. The wall is merely translucent at that frequency. That’s how these signals can be used to see through walls. Likewise, water, which is generally clear at visible light wavelengths will be quite opaque to microwave and absorb quite a lot of it. On rainy days, the humidity in the air could cause increased scattering of point-to-point Wi-Fi links to the point you might be able to see the link itself, and see it bend (or refract) in the air. Cool thought!
But then, it’s still not as simple as that, because reflection plays havoc with Wi-Fi signals (and radio signals in general). It causes a phenomenon known as multipath – basically due to superposition, waves can add in both constructive and destructive manners. So the strength of the signal at one point can be drastically different to the strength of the signal just a few centimeters away. If you visualized the intensity in 3D as a graph, it would look like a mountain range – signal strength goes up and down like mad. And if you moved reflective objects (metal ones, for example), the pattern would change. It’s kind of like the reason why if your food isn’t spinning in a microwave, you get hot spots and cold spots!
Wi-Fi Temporal Behaviour
Another thing completely ignored by the visualization is the temporal behaviour (i.e. behaviour with respect to time) of the Wi-Fi signal.
If you had a spectrum analyzer or a radio, you could easily visualize this – but I will describe several “features” of the Wi-Fi system which I think is most important to understanding how Wi-Fi would look if we could perceive it.
The first thing to look at is the behaviour of the Access Point (or stations participating in an Ad-Hoc network). All access points send out periodic beacon frames as part of their operation. These are short packets that say “I’m here, this is my MAC and SSID” (oversimplified, but lets keep it that way). These are sent at 100ms intervals (or 200ms intervals in some Belkin routers, but is customizable). The beacon frame length varies, but is generally transmitted at the lowest rate to ensure compatibility with all stations and that it can be received at the furthest distance the AP is expected to serve. The beacon frames would probably appear as very short flashes which repeat at 10Hz – kind of light a flickering but dim light. Our eyes are actually not capable of perceiving very short flashes of light – this is aptly demonstrated by the use of Pulse-Width Modulation for LEDs as a method of dimming – it merely runs the LEDs at full power, but for a fraction of the on time (duty cycle).
Depending on the network loading and the technology in use – the AP could just be idle and appear as a flickering dim light, or it could light up more of the time as it is sending data. In-between sends, the AP would stop sending and go dark, as the client card sends its acknowledgement back. So it’s like as if the light is “dancing back and forth” between AP and client – although very very quickly. Working at machine speeds, it’s hard to figure out what’s going on, so it might just appear like both are “lit up” to varying brightnesses.
The pattern of “lighting up” would be loosely related to what an Ethernet port’s activity indicator LED would behave like.
The actual colour of the lighting will probably be perceived as a single colour, almost greeny-brown or thereabouts. The reason is that the Wi-Fi signal is coded as a frequency hopping spread spectrum signal – so the colour is actually changing very quickly (it stays still for the grand total of 4 microseconds at each “colour”). That being said, it should be easy to pick out which channel the access point is running on just by looking at its “median” colour.
If we could slow down time or perceive shorter flashes, then we would be able to perceive more characteristics of the signal – say 22Mhz/20Mhz single-width modes versus 40Mhz double-width modes by seeing it “hop” between a wider range of colours. We would also see the preamble which is sent before each and every frame that gives time for the transmitter to get to power and for receivers to synchronize to the signal. We would see it hop colours slowly initially – this is because the packet header is sent at the slowest basic rate, then it would speed up and even change hopping “tendencies”. One thing I don’t think our eyes would be able to distinguish is the phase of the signal itself – so we might see a short burst of multiple colours in-between symbols when it’s changing phase, and a solid colour when that symbol is being “sent”.
The modulation isn’t as simple as “on or off” or “peak and trough” as the original article seems to imply. In fact, it is a phase encoding/quadrature keying of data onto the carrier.
And with modern Wi-Fi beamforming techniques, you might be able to see just how “bands” of signals (similar to pie shaped wedges) would be formed that have stronger signals towards the client, as opposed to radiating the signal all the way around. That way, you can (if the air is moist enough or your eyes sensitive enough, or if there are enough walls that show the path) see exactly who the AP is talking to at any given moment.
Maybe I’ve confused you all – maybe you all prefer the pretty picture, but the reality is that things are never as simple as they first appear. To be, even partly more true to the reality, complicates the situation greatly.
In fact, there are tools which help you visualize the Wi-Fi signal – but in different ways. Some site survey tools give you the ability to produce a “heat map” style display which shows the strength as a colour gradient on a map. There are spectrum analyzer dongles, which produce gradient displays showing signal intensity, versus frequency and time, at a particular location. None of these really capture the behaviour completely, in one fell swoop.
I hope my post has given you some ideas to use to imagine just how we would see Wi-Fi signals – if we could.