Another topic to consider is noise and interference. Ubiquiti products include a spectrum analyzer, known as AirView, that can show you a graphic of spectrum users detectable by the device's antenna. Usage of this tool will be covered later in the book. For now, we simply need to consider the relationship between signal and noise.
The link-budgets above assume that you are the only radio user in the region, and that there is nothing else affecting the portion of spectrum you wish to use. Unless you are building a network in a truly remote portion of the arctic, it is very unlikely that this is true (and even then, the water-levels in the air might attenuate your signal!). Therefore, it is important to use AirView and similar planning tools to try to select the least-noisy channel available for a given link. Even then, the radio will always encounter some noise. This will be expressed as a “Noise Floor” in the radio's user-interface. On new equipment, it is not unusual to see a reported floor in the -100 dB range; on older equipment, -95 dB was considered excellent. Noise ratings in this range are around the best you will see – much better than that, and you are seeing the results of the antenna/transmitters heating and internal circuitry! Sometimes, you will run into a horrible noise floor, maybe as poor as -80 dB (or worse; the worst I've personally seen is -65 dB).
To understand how this affects radio performance, you need to think about the receiver. The radio is performing multiple “fast Fourier transforms”, a mathematical process used to extract signal from noisy analog environments. With a noise floor of -95, and a signal of -75, there is a comfortable 20 dB difference in volume between background noise and signal; hence, it is relatively easy to extract the data you need. With a noise floor of -80 dB, and a signal of -75 dB, there is only a 5 dB difference between the two. Just like being in a noisy room with everyone talking at once, the receiver has a much harder time determining what sounds constitute signal (what you're trying to say), and what constitutes background noise.
The observant reader will notice that Noise Floor is another method of representing Signal-to-Noise Ratio, used extensively by Shannon’s Law (see page 24). This is another way of expressing the same basic truth: you need to have a clean signal in order to reliably transmit at high data-rates.
A good rule-of-thumb is to aim for at least 20 dB in separation between signal and noise. However, you can literally never have too little noise – the better the separation between the two, the better your equipment can perform.
One thing to note now (and explore in depth later in the book) is that a really common problem for WISPs is “self-interference”, or as I prefer to call it, “self-inflicted interference.” Two radios on a tower, close together, can emit noise that partially deafens their immediate neighbors. This is particularly true if the radios are on closely proximate channels, but in some cases at very short-range a radio can impair a neighboring radio on a relatively distant frequency also. It will be discussed in depth later, but I wholeheartedly recommend investing in “RF Armor” (from rfarmor.com) for any radio that will be mounted nearby another. It works wonders, and can render much of the problem of self-interference moot.« Chapter 4: Fresnel Zones Up To Contents Chapter 4: Real World Path Loss Estimation »
If you like this book, please consider making a donation to the author, Herbert Wolverson. If there is sufficient interest in this e-book, we'll post more chapters.