WPA2 EAP/PSK question

[jamezdin]

I'm trying to decide between WPA2 with PSK or WPA2 with EAP. Although this is just a SOHO setup I guess I would still like to use a more secure setting.

I have three questions:

1) Is there some kind of PSK encryption handshake required before PSK (user/password) info is exchanged?

2) What are the security benefits of EAP besides easier user access management?

3) Is MAC address encrypted in WPA2 traffic?

Thanks in advance.

[M/Q]

This link is a great white paper on WPA2,and you can get a better and more detailed description of what the process is.

http://www.wi-fi.org/membersonly/getfile.asp?f=WFA_02_27_05_WPA_WPA2_White_Paper.pd f

Your first question is not exactly accurate as the PSK information is not exchanged, that is the whole point of having to configure it in each device, to eliminate that issue.

The second question is more of an opinion on my part. I am not sure if there is any difference in security between PSK or EAP. It is just the initial authentication process. EAP has the obvious advantage of not having to configure the key on individual devices, but it has significant management overhead as well as usually another computer/device in the mix as the RADIUS server. I personally feel that PSK if there is not a significant number devices would be sufficient. As PSK and EAP are only used in the initial authentication process as far as I understand. Then after that authentication a 4 way handshake takes place between the AP and the client to establish the AES encryption keys.

Third question is somewhat nebulous on my part. I am fairly certain that during the initial AP/Client handshake to associate the MAC addr is sniffable if you will. Then after the PSK or EAP authentication it is encrypted. If this is in-correct, I would appreciate hearing about it.

[oliner]

This link is a great white paper on WPA2,and you can get a better and more detailed description of what the process is.

http://www.wi-fi.org/membersonly/getfile.asp?f=WFA_02_27_05_WPA_WPA2_White_Paper.pd f

Your first question is not exactly accurate as the PSK information is not exchanged, that is the whole point of having to configure it in each device, to eliminate that issue.

The second question is more of an opinion on my part. I am not sure if there is any difference in security between PSK or EAP. It is just the initial authentication process. EAP has the obvious advantage of not having to configure the key on individual devices, but it has significant management overhead as well as usually another computer/device in the mix as the RADIUS server. I personally feel that PSK if there is not a significant number devices would be sufficient. As PSK and EAP are only used in the initial authentication process as far as I understand. Then after that authentication a 4 way handshake takes place between the AP and the client to establish the AES encryption keys.

Third question is somewhat nebulous on my part. I am fairly certain that during the initial AP/Client handshake to associate the MAC addr is sniffable if you will. Then after the PSK or EAP authentication it is encrypted. If this is in-correct, I would appreciate hearing about it.
Interesting!
I have another question. If we need an authentification for machines(machineto machine world) what would be the most appropiate one and why?

Would it be EAP or more AES?

Thanks in advance

[M/Q]

This question might be better answered if we knew more about the situation. My guess this would be better handled by a network operating system. Are you using MS products? Whether this is wireless or wired machine to machine association and authentication is usually taken care of by the NOS. If it is MS than you would use Active Directory to authenticate machines.

If you would explain the situation in a bit more detail maybe we can get more granular with the answers.

[jamezdin]

Thanks for answering my (first) question. That white paper is awesome.

I'm confused about MAC addresses being encrypted after inital authentication. I though that MAC address in OSI model is infront of the encrypted data. Is that correct? How can networked computer recognize WPA2 traffic if MAC address is encrypted -- I mean how it knows which packets are meant for it?

Thanks again!

[M/Q]

That whole WiFi Alliance website is full of good info, I use it all the time.

As for the MAC addr question, I have been thinking about that all day. Do you have a MS computer with a wireless card in it. If so we could determine this for sure if you would get a packet sniffer like Ethereal and load it onto the computer. Then hopefully Ethereal and the wireless adapter will play nice together. I would but my Apple's wireless adapter does not want to go into promiscuous mode with Ethereal working.

I maybe back pedaling on my assumptions of the encryption, you can see I was hedging my bet on my first post. I will research it further. I guess it was something that I never was concerned about before now. On a quest now.

EDIT:

I think I understand where I had read about where the source and destination MAC address are encrypted. Moving forward, the 802.11i (and by extension WPA2) standards call for the use of Counter Mode with Cipher Block Chaining Message Authentication Code Protocol (CCMP), which specifies use of CCM with the AES block cipher. CCM is a general-purpose cipher mode that does not specify the block cipher to use. The CBC-MAC portion provides data integrity and authentication while the counter mode does the actual encryption, protecting the data from eavesdroppers.

When a packet is encrypted using CCMP, a number of data fields are added. The first field is the message integrity code (MIC), which is appended to the data. The MIC includes the hardware MAC addresses of the source and destination; this data essentially acts as a very strong cryptographically secure hashing function, which prevents man-in-the-middle attacks and other risks. The data and the MIC are then encrypted using the appropriate encryption key.

The packet is then modified with a data header. The first portion of data contained in the packet is the IV and key ID (4 bytes), which is needed to identify the encryption key used to encrypt the packet. At this point an extended IV (4 bytes) is attached to the packet. This field and the IV with key ID field are not encrypted, as the remote end must identify which key was used to encrypt a packet and the packet's sequence number. The first IV ensures that data is ordered properly. The rest of the packet contains the encrypted payload of data and MIC.

So, you were correct in your assertion that the source and destination MAC address for that hop are not encrypted, which only makes sense. The key here is for that hop. The original (first hop source) MAC address and the final destination (last hop destination) MAC address are used as a hash to check data integrity when the packet gets to the final destination. So the un-encrypted MAC source and destination address for each hop is similar to a link in a chain, where the next link/hop has no idea what the previous link/hop source MAC address was.