There's a thorough review of NetStumbler at findmysoft.com. Their rating is "Very Good" - thanks!

Congratulations to Helium Networks on the launch of their Wireless Recon product. Wireless Recon includes dedicated mapping hardware and software that combine to give you speedy and accurate site surveys. If you've used NetStumbler, then the SiteSense client may look rather familiar to you.
Read the full press release here.

Making effective use of available materials: you can build a cantenna from plastic water bottles, window screen mesh, cheap coax cable, and valve stems.

I've put up the "most reported to work" cards at stumbler.net/compat. This is based on a year of reports from users that were kind enough to send them in.

Earlier I read today's hack a day post about GM Onstar. Of course I need to get hold of one so the soon-to-be-released 0.4.1 can include support for it. If you have one (or possibly a Motorola Oncore board) and would like it to go to a good home, please contact me. (Just the VIU or VCIM box please, not the whole car)

Brad Templeton suggests extending beacons to include additional local information (lat/long, address, factoids, directions, hotspot info, etc). This is a great idea; I look forward to adding support for it into NetStumbler.

Everybody and their dog seems to be weighing in about Municipal Wi-Fi. Of course, they all have axes to grind and are taking the sides you would expect: the incumbent ISPs, telcos and cable companies are against it (sometimes in the form of poorly veiled "grassroots" organizations), while the community groups, potential users and equipment manufacturers are for it. No surprises, nothing to see here, though the outcome might be interesting.

I get asked this question rather too often, so I'm posting my short answer here. The answer is rather more complex than it ought to be, and depends on a huge number of factors.

The most important is the receive sensitivity of your equipment. Many manufacturers fail to publish this data, but those that do will generally rate their radios by dBm at various data rates. As an example, let us take the venerable ORiNOCO Gold 802.11b "Classic" card. Its receive sensitivity is:

• -94 dBm at 1 Mbps
  
• -91 dBm at 2 Mbps
  
• -87 dBm at 5.5 Mbps
  
• -82 dBm at 11 Mbps
  

In theory this means, in order to operate at 11 Mbps, this card must be consistently receiving a minimum signal level of -82 dBm. Any less and it is likely to drop to one of the lower rates; if you get as low as -94 dBm then the connection may drop altogether. As I mentioned before, many manufacturers do not quote their receive sensitiviy for their adapters; if you have one of these, I suggest picking a conservative figure such as -76dBm at 11 Mbps, which is the number for the Belkin F5D6020.
The signal level you receive in an unobstructed environment depends on the transmitter power, the gain of the two antennas involved, and the distance between them, as well as any loss between the antenna and the radio at each end.

In practice, radio waves behave unpredictably in a number of ways. First, the signal will fade out due to multipath effects (radio waves that bounce off objects and increase or decrease the signal that you receive). The further the receiver is from the transmitter, and the more objects between them, the higher this effect will be. Walls, people, electronic equipment, rain/snow/ice/fog are all quite effective at decreasing your signal level. In a typical home or small office environment without too many obstructions, a 10dB variation in signal level is quite normal. So, if you are looking at a NetStumbler scan and the signal is consistently around -65 dBm, it could drop to -75 dBm when somebody comes over to talk to you.

Summary so far:
(Received signal) = (transmit power) - (loss between transmitter and antenna) + (transmit antenna gain) - (path loss) - (multipath and obstruction loss) + (receive antenna gain) - (loss between antenna and receiver)
In order to operate, (received signal) must be greater than (receiver sensitivity).

Another factor is noise. This is "background" radio-frequency junk that your receiver can "hear" but needs to reject. Sources of noise include other wireless networks, cordless phones, microwave ovens, radio hams, medical equipment, Like other radio phenomena, noise may be highly variable. Many wireless network adapters do not report noise, so if you're using NetStumbler with them then you can't even tell how much noise you have in your environment. A typical urban location these days might have an average noise level around -95 dBm. When you switch on the microwave oven or take a call on your 2.4GHz phone, this value will increase. I've seen a 2.4GHz phone produce -50 dBm of noise, which is enough to saturate some Wi-Fi radios and thus kill their connection completely.

Let's take these concepts and combine them. In order to operate, the actual signal level at your receiver needs to be higher than the noise level. The actual signal level varies depending on signal fade, so if you measured -75 dBm one day, it might drop to -85 dBm occasionally. On most radios this is sufficient to make it drop to a lower data rate, and on some it will cause the connection to drop altogether. Likewise your background noise might be around -98 dBm, but then your neighbor takes a call on her cordless phone and it jumps to -78 dBm. With multipath effects, this is sufficient to make your connection drop randomly.

My conclusion, therefore, is:
Q: What signal level should I consider usable for a good wireless link?
A: Depends on your equipment and your environment.

Softpedia has awarded NetStumbler 5 stars and a Softpedia Pick Award.

According to a very well written and thorough article in the Virginia Journal Of Law & Technology, what we've been saying for over 3 years has been determined to be true: WarDriving is not a crime.