Unfortunately, there are a few holes and flaws in the theory/opinions.
Not going to get into it full boar as it is not my thread to clean-up....
My credentials, I have been a Radio & IT tech for over 15 years so I have a pretty good understanding of the issues, pitfalls and Pro's.
OS's: XP is good but needs to be definitely SP2 or later to allow for half decent encryption. For best results, always use Windows to manage the device/connection. Third-party apps quickly add other BS and headaches when things stop working properly.
Vista, what a lemon of an OS. It is a mongrel to use but more importantly, ensure that you use good quality hardware that is well supported, avoid stuff like D-Link and Netgear as the drivers are nothing special like the hardware.
Win7, a much better OS than the rest but same again, you need to familiarise yourself with it to get best results.
Linux/MAC, can have mixed results but again, get to know your OS and it's way of doing things.
Hardware and modes:There are four common modes of operation and have different benefits and pitfalls. These A, B, G and of course N as well as the variants of N.
802.11a was an early standard using the 5.8GHz spectrum that at that stage was not overly flogged out. The down sides are that 5.8GHz is very easy to block but also very easy to reflect of bounce. This would be good for N-MIMO but not for the old standard that didn't handle multi-path time-shifted signals.
802.11b was also and early puppy but used the 2.4GHz band. At it's first appearance, the 2.4GHz was still reasonably clean and had some advantages over 5.8GHz. Still easy to block or bounce but not as much as the higher microwave frequencies. However, B does work well in noisy environments, don't discount it because it's a little older.
802.11g, well this mould be one of the best known standards and does work well when tweaked properly. Also in the 2.4GHz band that is now rather crowded but other devices.
802.11n, this is where the magic is if you are smart or lucky enough. There several classes of N standard and these are spread over both the 2.4 and 5.8GHz bands as well MIMO. MIMO stands for Multiple In, Multiple Out. What this does is that your devices (if full MIMO compliant) will have multiple radio stages and antennas to allow up to three simultaneous connections however, these connections can be of different time alignments and signal strengths thus dealing with multiple reflections of different arrival times that would affect all previous forms.
Hardware plays a big part in wireless, if you buy cheap then don't expect much. Always aim at a well know and supported product. Also look for a product that either support the features you want but also supports half-decent firmware's. A good example is the mighty Linksys WRT54G/GL, these are basic and have been around for a bloody long time but they work and work very well, especially with flashed with something like Tofu, DD-WRT or Tomato. All are feature rich but also support RF power level control as well as QoS.
Avoid external antennas that have low-grade and/or long coax leads. There is no coax that is good for microwave, all leak like ****, just in various levels. Good stuff is over 10mm thick and hard to work with, better is over 1 inch or bigger. Real men use LDF (hard-draw) or wave-guide.
Don't use power boosters or eBay ****. Most of this stuff is illegal and just makes things worse, especially for others.
Antennas need to be designed for the task and band, any old bit of wire just won't do. The four main types are Yagi, Omni-directional, grid-pack reflector and phased-arrays (better known as panel antennas).
Yagi's are good for long distance weak but clean signals. They are not much good for scattered signals (obstructions in the way of LoS (Line of Sight))
Phased-arrays are best for both longer ranges but more so if the signal is scattered.
Omni's are for when you want 360 degrees of RF pattern, common for the AP. The catch is not to use a design that has too much gain. Bigger sticks actually are detrimental as they become lossy after 6Db.
Grid-packs are for high-gain directional work over long distances, especially with obstructions.
There other things in the home-brew departments like "Can-tennas" and reflectors but these in most cases a WOFTAM if you don't know what you are doing. Can-tennas are best used for inflating an e-penis rather than being functional.
Spectrum:The recommendation to use one channel or another is just plain silly and obvious. Each setup/location is different and has different requirements. What does need to be done is to keep your WiFi channels well away from any others in the area. This is not easy for some as it is a required feature to "Scan or Survey" the band. Other things that can be used are the likes of "NetStumbler" that can give you a good representation of what's around you RF wise. The magic advice is as follows, you need to be at least 3 channels away from the others and if there are multiple channels being used in the area, then you need to be far away from the stronger ones, weaker will cause less issues.
The reason why is what's knows and frequency bandwidth. Each channel in the WiFi band is actually 3 channels wide, not 1. I won't explain it in detail here as it is long-winded and best for those of us savy in RF.
Testing is always needed and the best and basic ways are to simply run a large file transfer while running several pings (CMD box) to both local and external addresses. This will soon see if there is congestion or noise in the area. High pings and dropped packets is good sign of high RF noise floor or poor signal/signal to noise ratio (SNR).
If you want some good reading, have a look at this Wiki, ity isn't too bad and gets the bulk out.
http://en.wikipedia.org/wiki/IEEE_802.11 As for any wireless, always use a cable if possible rather than wireless. If wireless is to be used then ALWAYS lock it down with at least WPA/WPA2 and do your homework first.
Cheers,
LB
