Forum

We know that in an 802.11 system, some STA???¡é?¡é?????¡é???¡és [ STAtions ???¡é?¡é?????¡é?€?? laptops etc ] can be very near the access point. But some can be far away. Some can be so far away that due to dynamic rate shifting, their data rates can drop down to a measly 1 Mbps [ In a well configured system we wouldn???¡é?¡é?????¡é???¡ét normally allow that, by disabling very low rates such as 1 and 2Mbps, due to a bunch of factors ]. However the spec says that we must cover that possible situation. We know that radio waves travel at 3 X 10 to the 8 m/s in a vacuum and almost that value in a room or outside. That sounds like a very fast rate, and indeed it is, but that time to go from the Access Point to the STA has to be taken into account [ even though it is small ]. How does the 802.11 spec take care of this ?

We are familiar with MAC addresses etc which let us know who is calling who, but how do we know that a radio transmission is incoming to our receiver, and not just a bunch of noise ???¡é?¡é?????¡­?¡°disguised???¡é?¡é???????? as a signal ?

The answers are quite interesting.

Dave

When an ACK frame is sent from an access point to a station which has just sent a frame, the AP sending the ACK waits a period of time known as the SIFS [ SHORT ???¡é?¡é?????¡é?€??Inter Frame Space ] before sending it. For 802.11a,g and n systems, this value is 16 microseconds.

Before a beacon is sent, the AP waits a period of time called a PIFS [ PCF Inter Frame Space ]. This PIFS is made up of two parts [ they???¡é?¡é?????¡é???¡ére all one continuous time, but the calculations in the receiver use a formula to calculate that time based on two parts ]:

1. The SIFS time period

2. What is called a ???¡é?¡é?????¡­?¡°slot time???¡é?¡é????????

For 802.11a, g and n systems, this slot time period is 9 microseconds.

Before data is sent, a station may wait for a DIFS [ DCF Inter Frame Space ], wait for a ???¡é?¡é?????¡­?¡°backoff???¡é?¡é???????? period and then send the data. This DIFS is made up of two parts:

1. The SIFS time period

2. Two slot times

We know that during the ???¡é?¡é?????¡­?¡°random???¡é?¡é???????? back-off period, the transmitter selects a value equal to a number of slots. That number is then multiplied by the slot time to give us a ???¡é?¡é?????¡­?¡°backoff???¡é?¡é???????? time value.

But how did they come up with that number of 9 microseconds ?

This is where propagation time comes in. Without going into all the details, the designers of 802.11 knew that certain minimum bit rates had to be supported. Also , certain minimum values of bit-error rate had to be supported. Knowing what sort of carrier to noise value was required to support this value, you can work backwards to get a maximum range for a particular signal. [ Actually the metric used is what is called the ???¡é?¡é?????¡­?¡°Energy Per Bit To Noise Density Ratio???¡é?¡é????????, but it is related to C/N ]. Practical measurements were then done to work out maximum cell sizes. Once that was done, they knew the maximum distance apart a station and an access point could be. Once they had that, they could work out the maximum time difference it would take for an RF signal to get to the furthest point in the link. Hence maximum propagation time value. Transmission takes place on a slot boundary.

This was the first factor in coming up with a value for the slot time.

The next thing was: ???¡é?¡é?????¡­?¡°How do we know that a signal being received is really an 802.11 signal and not just a burst of noise ????¡é?¡é???????? The way they did this was to include certain ???¡é?¡é?????¡­?¡°markers???¡é?¡é????????at the physical layer. The physical layer is made up of two parts:

1. The PLCP

2. The PMD

Without going into the definitions of the above terms, they basically do this:

The PLCP adds some special headers with special fields. Some of these fields tell the receiver what speed the payload MAC frame is being transmitted at, etc.

The PMD is where the actual modulation takes place. This could be Infra-Red in VERY old 802.11 systems or PSK or QAM in more modern ones.



There are two special fields right at the beginning of every 802.11 transmission. These are called the ???¡é?¡é?????¡­?¡°Pre-amble???¡é?¡é???????? and the ???¡é?¡é?????¡­?¡°Start-Of Frame Delimiter???¡é?¡é????????. The pre-amble has a particular pattern [ in old FHSS systems it was an alternating zero ???¡é?¡é?????¡é?€?? one pattern ]. The receiver looks for this pattern [ the pattern was chosen for many reasons, one of them being to NOT look like noise, which for the most part is random ] and when it recognizes it, it says ???¡é?¡é?????¡­?¡°Ah hah, an 802.11 frame is coming in. Right after that, we have a SOF de-limiter. This basically says ???¡é?¡é?????¡­?¡°OK, you have just found out that an 802.11 frame is incoming. Right after you see the end of MY special pattern [ different from the pre-amble ] all sorts of useful stuff will be coming ???¡é?¡é?????¡é?€?? info on payload data rates, some info on the length of the data frame etc ???¡é?¡é?????¡­?¡° Before we even get to MAC frames and all that sending station MAC and receiving station MAC stuff, we MUST process the PHY layer information. This takes time. This time period was the second major element in determining the slot time.

So, slot time takes into account propagation delays and enough time for a transmitting station???¡é?¡é?????¡é???¡és preamble to be to be detected by other stations no matter where in the cell they may be.


Notes:
The term ???¡é?¡é?????¡­?¡°Signal to noise???¡é?¡é???????? should really be called ???¡é?¡é?????¡­?¡°carrier to noise???¡é?¡é????????, but S/N is used in nearly all documentation ???¡é?¡é?????¡é?€?? signal to noise comes from old analog days and referred to say, audio signal to noise AFTER the demodulator. BEFORE the demod we have carrier to noise. However ???¡é?¡é?????¡­?¡°everybody???¡é?¡é???????? calls it S/N, so I will too.

When you have systems using different modulation values etc, how do you compare apples with pears ? You use what is called the Eb/No value. More about that in another posting. It is the only true way to compare one system???¡é?¡é?????¡é???¡és demodulation capabilities with another.

Dave

Bye the way, the main purpose of SIFS, PIFS, DIFS is to prioritize access to the medium for different frame types. For example, we like to try and send beacons out at regular intervals [ doesn't always happen - sometimes they have to "wait a wee bit" whilst other frames are transmitting ], but for the most part, we would like to have them transmit before data frames, say. By making the PIFS value smaller than the DIFS value, we can help accomplish this. ACK frames used to have the smallest value of the "IFS's", but there are now other little horrors creeping in like the RIFS.

Dave