14

I am new to wireless LAN. I see that it has four addresses in the frame header. In case of a Wired Ethernet, there are only two MAC addresses, the source and the destination (which is essentially the address of the next hop).

But in case of wireless LAN, why do we always need the end point's address? Isn't this task essentially for the IP layer?

What is the need of the extra source and destination addresses, apart from the transmitter and receiver addresses?

Matthias Braun
  • 165
  • 6
Ankur Bhatia
  • 243
  • 1
  • 2
  • 5

1 Answers1

19

The distinction to understand is that while an 802.11 device is transmitting to a receiving device, either one (or both) of these devices may not be the actual source or destination of the L2 traffic. So this can create situations where you need four different addresses:

  • Transmitter Address (TA)
  • Receiver Address (RA)
  • Source Address (SA)
  • Destination Address (DA)

The BSSID of the wireless network may be the TA or RA, but it can also be an identifier to associate traffic to a BSS. Unless the access point (AP) uses the same MAC address for the management interface and the BSSID, you generally won't have any traffic where the BSSID is either the SA or DA.

The use the address fields in an 802.11 header is tied to the To/From DS (distribution system) flags. Depending on how the To/From DS flags are set, this determines which of the four fields are required and what information each provides.

Here are a couple quick summary tables that I took from an IEEE document:

IEEE table of distribution system flags with meaning explained

IEEE table of distribution system flags with different addresses

Let me add a couple of images to provide examples taken from 802.11 Wireless Networks: The Definitive Guide by O'Reilly Press:

illustration 1 of client, access point, and server with their source, transmitter, receiver, and destination addresses

illustration 2 of client, access points, and server with their source, transmitter, receiver, and destination addresses

Matthias Braun
  • 165
  • 6
YLearn
  • 27,141
  • 5
  • 59
  • 128
  • 2
    Thank you Ylearn. The mistake I was doing in understanding this concept was considering the other clients as a separate network. They are actually in the same link, hence we need the mac address. And in addition to the source and destination, we now have an intermediate ie the access point, which is also in the same link, so we need the mac address. And similarly for the scenario where we have more than 1 access point we need all 4 addresses. Thanks a lot again for the explanation. – Ankur Bhatia Dec 11 '15 at 17:30
  • 3
    @AnkurBhatia, in normal Ethernet devices are connected by wires. This is not the case in 802.11 so one way to think of it is that the TA/RA are identifying a "virtual wire" connecting the two devices. The SA/DA will only match the TA/RA when the device sending/receiving the frame is also the device sending/receiving the RF transmission. – YLearn Dec 11 '15 at 22:00