Last Post: January 1, 2006:
I am wondering where EIFS (Extended Interframe Space) fits in relation to other interframe spacing. I know that in order from shortest wait time to longest wait time it is SIFS, PIFS, and then DIFS but where does EIFS fit?
-Reference CWNA guide pg. 368-371
EIFS is the longest of the interframe spaces. A PIFS is equal to a SIFS + Slot Time (Slot Times are defined for each PHY). A DIFS equals a SIFS + 2 Slot Times. The formula to calculate an EIFS is much longer but it includes a SIFS and a DIFS. From 802.11-1999 "The EIFS is derived from the SIFS and the DIFS and the length of time it takes to transmit an ACK frame at 1 Mbit/s by the following equation:
EIFS = aSIFSTime + (8 x ACKSize) + aPreambleLength + aPLCPHeaderLngth+ DIFS"
EIFS is used in DCF after a station senses an idle medium following reception of an erroneous frame- one in which the MAC frame is incomplete or the MAC frame has an incorrect FCS. The EIFS allows enough time for another station to acknowledge the frame (it may have been received correctly at the intended station) before this station begins transmission.
Very well said. Thanks. /criss
One way I've found to help people understand EIFS is the following.
When a station receives a corrupted frame, it has no way of knowing which part of the frame was corrupted. Therefore, it must assume that the frame never made it to its intended receiver. If the frame never made it to the intended receiver, that means that a Retry must happen. The EIFS forces stations to backoff (not transmit) for an extended period of time while a Retry happens after a corrupted frame.
So the other stations back off, not the offending station that transmitted the corrupt frame. I had it backwards. Thanks
Remember that when a station transmits, it cannot hear itself. That's why wireless is collision avoidance instead of collision detection.
After which timers, a transmitting Station will do retransmission ??
The transmitter of a unicast addressed frame expects to receive an Acknowledgment frame (ACK) within a Short Inter Frame Space of time (SIFS) after ending the transmission. If it does the transmission event is finished and that frame is cleared from the transmission queue. If it does not, and the retry limit has not been exceeded, then it contends again for the medium with all the attendant backoff counters and timers.
The transmitter of a multicast addressed frame expects no ACK and, after transmitting, clears that frame from the transmission queue and moves on.
I hope this helps. Can you add your location to your forum profile? Thanks. /criss
IEEE 802.3 Ethernet uses collision detection and yet Ethernet stations do not hear themselves when they transmit.
Ok, they do sense the medium for either current (coax) or incoming signal from the immediate link partner (twisted pair or fiber when used in half duplex).
Ok, when a coax station senses current it is a composite of the currents imposed by all overlapping carriers, so in that sense they hear their own carrier but not their own signal.
But there is an other reason for why 802.11 uses collision avoidance rather than detection. Even if an 802.11 station was built to hear its own outgoing signals along with other incoming signals that station would not be able to conclude from those signals what is happening in the rest of the medium. Ethernet can, so long as the minimum frame length and maximum cable length rules are adhered to. So a half duplex Ethernet station can stop a transmission in its tracks and start over in an instant if it gets "any" incoming signal as described above. And if it gets no incoming signal then it can logically conclude that "all" stations on the medium have correctly received the frame.
I hope this helps. Thanks. /criss
Thanks Criss - I guess what I should have said, is that an 802.11 device cannot hear anything when it is transmitting, thereby not being able to "detect" a collision. Thanks for the clarification!