Last Post: June 7, 2006:
IEEE 802.11 WLAN is designed to mimic IEEE 802.3 Ethernet. Software operating from the LLC layer and up is largely oblivious to what is happening below LLC. Also by design WLAN and Ethernet data-links readily bridge together into what appear to LLC and above as a single data-link or catenet. When used with TCP/IP each catenet is naturally a single IP subnet.
Two Ethernet stations on a single Ethernet exchange frames directly and require only two MAC addresses. Two WLAN client stations on a single infrastructure WLAN exchange frames indirectly and require the MAC addresses of any intermediate access points. These addressed intermediate devices coordinate access to the shared RF medium, extend RF range between WLAN client stations, and provide a portal to other media. Each client station fixates on its one access point and exchanges frames with it alone, regardless of the ultimate MAC destination for a frame payload.
The WLAN designers gave us the following options for indirection within a single WLAN.
1. Independent BSS (IBSS). All client stations communicate directly in a mesh with no indirection. However a third MAC address is used in data frames to identify the IBSS to which the frame exchange belongs. These WLANs are relatively rare.
2. Single access point (BSS). All client stations communicate indirectly by way of their one access point and require three MAC addresses to exchange data. Most WLANs use this paradigm.
3. Multiple access points that use the same RF medium to relay as their client stations use (ESS). All client stations communicate indirectly as above, but the access points implement a frame payload relay between themselves. This is the relatively inefficient Wireless Distribution System (WDS) that famously requires four MAC addresses to relay data between access points. These WLANs are relatively rare.
Multiple access points may use some other medium than the one used by their clients to relay their clients' frame payloads between themselves. While the standard does not require this other medium to be a bridged Ethernet, this is what the designers had in mind and is what the vendors have built. Regardless of how many WLANs and Ethernets are bridged together, LLC software on all attached stations operate as though they were on a single IEEE data-link. IEEE 802.11 defines multiple bridged WLANs to be a single ESS as well.
A typical WLAN client is configured with the IP address of its default gateway which by definition must be on the same IP subnet as the client. ARP brings back a MAC address for the default gateway. The WLAN client does not care a whit whether that MAC address is on its own WLAN data-link segment or on another data-link segment, wired or wireless, of a larger catenet. If WLAN clients were required to use an IP address of their associated access points as their default gateways, such as with home "wireless NAT routers", IP addressing in a corporate WLAN would be unmanageable and roaming between access points would be virtually impossible.
I hope this helps. Thanks. /criss
Thanks responding to these post Criss, you are Simply Phenominal! I am nowhere near where I desire to be and look for these comments just to understand the flow of communications at that level of details. Your contributions have help me in more ways that just certification. It helps me logically see the standards and approach them from the angle of the IEEE.
Thanks a lot for the informative post. Like the earlier guy said, you indeed are a phenomenal resource!