SSID

From whom are you getting permission? I didn't know there were rules in such matters.

It should be in the IEEE 802.11 specification.

From IEEE 802.11-1999 7.3.2.1

The SSID element indicates the identity of an ESS or IBSS. See Figure 35.

The length of the SSID information field is between 0 and 32 octets. A 0 length information field indicates the broadcast SSID.

That's a bit misleading as the SSID is sometimes null terminated leaving only 31 characters available. Some firmware versions screw up if you use the full 32 characters.

The characters must also be printable, so no control characters are allowed. I'm fairly sure that a leading space is also not allowed, but there doesn't seem to be anything specified.

Also: The following six characters are not allowed: ?, ", $, [, \, ], and +. In addition, the following three characters cannot be the first character: !, #, and ;.

From a previous posting:

Spaces are just fine. Some utilities will blow up if you do that, but most access points handle it just fine.

Maximum humor, keep it clean, forget about bell characters, and no nulls.

Give? I *SELL* permission. Pay me.

Specification writing is the art of micromanagement. There is a rule for everything. If not, one will be produced. Sometimes, there are even more than one rule for everything. Without rules, there would be nothing for me to break.

Tested with my Cisco access point with latest firmware

Does not allow trailing spaces

Does allow leading spaces, tested and it works with XP and FreeBSD system

Again tested with Cisco they say only + ] / " TAB and trailing space is invalid

Tried ? and it works

Tried leading ! and it works

Interesting! When you say it works I presume you mean you can connect and transfer data rather than just see the SSID.

From the Cisco info. I have seen they are quite clear that those characters are not to be used.

Interesting. It appears that in the absence of "specific and detailed" valid SSID characters vendors have made up their own restrictions. I looked at "ANSI/IEEE Std 802.11, 1999 Edition (R2003)" and could not find any restrictions to the characters it might contain, only that it is an octet string of size 0-32. I admit my look was very cursive and I may have missed something, but it appears IEEE have dealt extensively with size and use of SSID and almost nothing on what it can contain.

In my own case I have two Linksys wireless routers (WTR54G and BEFW11S4) with SSIDs of the type #naaaaaa (e.g. #5window) and they both work with XP and Kubuntu.

Question: If the standards do not specify restrictions and a vendor applies restrictions, is that a violation of the standards?

Plan9 hath wroth:

There are bigger holes in the 802.11 specs than a failure to specify a char or string field type. Nobody ever considered the possibility that users would want to intentionally obscure the SSID, hide the SSID, or use it as any form of security. The previous standard was published in 1997 and I vaguely recall (and am too lazy to research) was originally inscribed in about 1994/1995. Now, think about where we were in computing 12 years ago and see if you could predict current applications. I think Windoze 3.1 and Xenix was the fashion. Linux hadn't even been started. Next, try to predict where we'll be 12 years in the future and write an airtight spec that includes all possible creative interpretations and mutiliations. Be sure to make it compatible with systems that do not yet exist, with future security considerations, with future government regulations, and with predicted fashion trends. If you can do that, I wanna buy the crystal ball you're using.

I lifted the SSID restrictions from the Cisco SSID Manager release notes. See 2nd paragraph:

My guess(tm) is that the various strictures are the result of various Cisco programs and shell scripts blowing up if confronted with these characters. That would certainly explain the strictures on the leading ! # ; which would probably blow up a Bourne Shell or Perl script. Leading and trailing spaces will cause problems with HTML in the configuration utilities. I found the problem with the null string delimiter which resulted in a 31 character SSID, instead of 32. As long as the management and configuration program handles the SSID as a string, there's going to be scripting problems.

As for violating 802.11 standards, methinks you should take that up with the Wi-Fi Alliance, which offers various certifications:

However, I don't think you're going to get their attention. They have successfully ignored the ASCII versus Hex WEP key concompatibility between different vendor implimentations for years. There are also some rather creative timing incompatibilities, such as those between Meru Networks and Cisco. The Wi-Fi Alliance apparently (my guess) has no interest in becoming an enforcement organization and is only interested in selling certifications.

Incidentally, I recently had a customer discover that it was possible to create a UserName in Vista that starts with a space. It was a typo error but it's driving me nuts because some utilities accept the leading space, while others do not. Directory names that start with a space are officially proscribed, but I'm still stuck with: c:\home\ user\ I can rename the user, but this customer already created a 2nd user with the same UserName, but without the leading space. Now, he's got files scattered all over the machine and randomly destributed between the two UserNames depending on how various utilities and programs handle the leading space. Worse, I can't move everything to a 3rd neutral UserName because he has multiple network authentication accounts all configured with the UserName.

So, go ahead. Try using a leading space and see what breaks.

Jeff Liebermann hath wroth:

Oops. So much for my photographic memory.

Linux was announced in 1991:

In 1995, SCO announce the end of Xenix support. OpenServer 5 had just been released. At least I got the Windoze 3.1 right.

Grumble...

What I quoted was from a pop up box generated by Cisco software on 1100 access point when I put a trailing space in the SSID

System Software Filename: c1100-k9w7-tar.123-8.JEC System Software Version: 12.3(8)JEC

Just tried SSID of ?$[\ no leading or trailing spaces, and it worked :) and yes transferred data. Cisco access point FreeBSD client.

Anyway enough silliness I guess there is no definitive answer.

Interestingly the problem I was trying to solve was exactly that. A client trying to logon to an access point and the software trimmed the leading space.

Not my access point I hasten to add. It wouldn't even occur to me to put in a leading space.

Anyway problem solved. Next.

~ Can anyone point me to a reference [rfc] that says what exactly is ~ permissible in a SSID ESSID ~ ~ like leading spaces what symbols etc ~ ~ TIA

Per the IEEE 802.11 standards, which are the standards that count, if you're looking for standards, there are no restrictions on the value of the octets in the SSID string. I.e. any value from 0 through

That said, specific implementations may have issues with particular octet values, for various reasons. For example, as some have noted, Cisco products will sometimes assert that they forbid certain characters. This is not because the use of those characters violates a *standard*, but because those characters cause problems for things like the IOS CLI parser and the HTML parser etc. Clients may exhibit similar limitations.

From an interoperability standpoint, you're better off sticking to straightforward characters in your SSIDs. It's not as though using exotic octet values in your SSID is going to confer any real security benefit.

Aaron