Forum

  • Hi,
    in 802.11bg the not overlapping channal usable to correctly manage roaming in a MCA are 1, 6, 11 (in Italy for example, because it depends from country); that's is right using 20Mhz Channels.
    With 802.11n, using 40Mhz channels to have high throughput and reach high data rates till 300 Mbs, what does it happen with channel re-usability concept? In this case, in 2,4 Ghz I would have just 1 channel available? If it's true, it means despite I will manually set my access point on 1, 6, 11 channels the clients will have troubles with roaming?
    Thanks.

  • If you use 40 mhz channels in 2.4 you can have only one "non overlapping" 40 mhz channel. It is not recommended to have 40 mhz channels in 2.4.

    Also to obtain 300 Mbs, your APs and client must not only use 40 mhz channels but be able to use 2 spatial streams, a short guard interval, and be located close enough with enough clarity to be using QAM 64 modulation.

  • You can use channel 1 and channel 5 to give as the center frequencies to give 2 x 20MHz =40Mhz. The 40MHz will cover channels 1 to 7. You can use channel 2 and channel 6 as the center frequencies which will cover channels 1 to 8 to get 40Mhz. You can also use channel 3 and channel 7 etc

    However as you can see by t henumber of channels the 40Mhz covers, if only channels 1 to 11 are allowed, there can only be one 40MHz channel in the 2.4GHz band.

    You will have to use an SCA to scale the solution for coverage, capacity etc. I only know of Meru that does an SCA.

  • Thanks for replies guys.
    that it was infact my doubts. I knew I could use just 1 channel wiidth 40 Mhz in 2,4 Ghz and to reach hight data throughput I need also greenfield, short guard interval and 2 spatial stream.
    Reading your answears, now I understand I cannot use high throughput in warehouses on 2,4 Ghz with MCA, just keeping 20 Mhz and 54 Mbs as maximum data rate.
    As I said, the mobile units cannot go over so, it's not a problem. I wanted give to my customer the best from product but that is not possible in the scenario I described.
    Thanks again.

  • Unless I'm missing something, why can't you have AP1 on channel 1 and AP2 on channel 11? Even though the BW is 40MHz, isn't there enough spacing for non-overlapping?
    Using InSSIDer, one AP would show 1+5, the other 11+15 with plenty of spacing.

    Warehouse setting needs high BW?

    Separate AP's in SCA, BW would be SHARED because of now co-channel inteference, BW would be reduced down to half or worse (unless you shrank the cells enough) so the AP's and Clients don't hear each other.

    I'm still learning myself -- does this sound correct?

  • There is no channel 15 as far as I am aware. 14 is max and only up to 11 allowed in most countries,

  • True -- I believe it is InSSIDer's way of letting you know its center frequency is where if there was a channel 15 (5MHz above channel 14), that that's where it would be.

  • Besides 40 MHz at 2.4 GHz being a bad idea for some of the reasons already stated, all clients I have seen default to 20 MHz channels at 2.4 GHz and this is changed in the driver. In other words, changing clients to use 40 MHz channels is not going to be easy in most enterprise environments, even if using 40 MHz channels was a good idea.

    Certainly Cisco does not support 40 MHz channels (at 2.4 GHz) which is certainly a good thing so all of this is academic if using their gear. I am unsure about other enterprise vendors but many SOHO Wi-Fi routers default to 40 MHz which is very lame.

  • Daniel

    RF interference is one of the most complex subjects in radio engineering. In many diagrams, we observe a representation of a particular frequency range as a simple rectangle. This is done for convenience in literature. In reality, this is impossible. To produce a signal which was exactly shaped like a ?box? would require what is known as a ?brick wall filter?. In other words, if we looked at the signal level going outwards ( left and right in terms of lower frequency and higher frequency from the center ) we would see ?signal, same signal, same signal?..noise floor?. When designing electronic filters using physical components, there are various mathematical design rules that show us that we cannot do that. In other words, we cannot have a ?90 degree dropoff? like a waterfall from the edge of the rectangle.

    In reality, Wi-Fi modulated waveforms have very complex shapes. There is a balance in design between what we call the ?roll-off rate? ( sort of how steep the slope is ) at the edges of the signal and factors such as how much ripple ( bouncing up and down ) is present at the ?top of the signal?the flattish bit? etc. These shapes are very complex. Sidebands are present. These sidebands obey mathematical rules similar-ish to the sidelobes that are present in antennas. When we look at a spectral mask required by the Wi-Fi alliance, it takes into account a number of these factors.

    Now imagine that we had a ?magic box? and that the output of a modulator was indeed a pure rectangular shape. Now, we take another ouput and have it overlap the original by say 10 %. Does that mean that the signal to noise ratio of the original wanted signal has now decreased by 10% ? The answer is that we cannot say that by simply looking at the simple ratio of overlap. The entire modualtion process is extremely complex. Certain calculations can be made to give rough estimates of the effect of interference. However, all manufacturers that I know of perform practical tests with a wanted carrier and one or more unwanted carriers.

    There is one major thing that is often misinterpreted when looking at channel layout diagrams. We see a number of channels with all the carriers having exactly the same power level. In other words, you could draw a line across the top of the carriers in the diagram, and they would all be in line. Real life is much different as we can observe on a spectrum analyzer. Depending upon distance etc away from the place that we are observing from, signals adjacent to our carrier of interest could easily be higher than or less than our carrier in terms of power. The effect of the same amount of overlap on a signal with a very low carrier to noise ratio can vary dramatically from that of a larger carrier. Sidebands may be more ?hidden in the noise? on one adjacent carrier than another. This can have a marked effect on the relative amount of interference produced.

    The overlapping channel plan used at 2.4 GHz in Wi-Fi is one of the most unusual in all of RF. The vast majority of all other channel plans used in microwave transmission, satellite transmission etc do not utilize that overlapping system. For example, if you are using channel 36 in the UNII ? 1 band operating in the 5GHz range, and another carrier appears at the adjacent channel number 40, then provided that both carriers conform to to a proper spectral mask from a responsible authority, such as the Wi-Fi alliance, then irrespective of the physical location of the adjacent carrier or the variation in signal level from movement etc, the carrier at channel 36 should not be affected. Guard bands play a vital role in this. They not allow for possible ?sideband creep?, but also for frequency drift of the actual carrier frequency itself. When we ?channel bond? at 5GHz, and another channel appears above or below, that channel should have no effect on the bonded channel setup, provided power levels etc are within regulatory compliance.

    2.4 GHz is a strange beast, as there are no actual defined guard bands per se. I won?t go into the details why.

    Now could you have two 40 MHz carriers at 2.4 GHz as mentioned. Yes you could under certain circumstances, but not using the frequency pair mentioned ( remember that we have 20 MHz below and above the center frequency for 40 MHZ operation ) but there are so many pitfalls that it is not suggested that this be done. Again, going back to the business of carrier levels on diagrams, where we see every carrier having the same level. That is rare in practice. It is possible that if we had a 40 MHz carrier at the lower side of the frequency band, and another higher, but at very low level due to distance for example, that we could get some sort of a service. The degree of overlap would be critical. However, from a practical point of view, this would not be a good solution. Error rates would be bouncing all over the place. Physically it is possible, but practically it?s a ?no-no?. In other words it?s not one of things where it is impossible to get any service at all, but the system stability etc would be terrible. If you have access to a spectrum analyzer ( and away from Wi-Fi neighbours etc !!.....not an easy one !!) you can actually set up two APs at 2.4 GHz 40 MHz and move one closer to the other and see the effects on performance. Again, this test needs to be done under controlled circumstances !! I?ve done it at home when I knew that all my neighbors were at work.

    http://en.wikipedia.org/wiki/U-NII

    http://www.home.agilent.com/agilent/editorial.jspx?ckey=332766&id=332766&lc=eng&cc=SG
    http://en.wikipedia.org/wiki/Sinc_filter

    Dave

  • [quote]For example, if you are using channel 36 in the UNII ? 1 band operating in the 5GHz range, and another carrier appears at the adjacent channel number 40, then provided that both carriers conform to to a proper spectral mask from a responsible authority, such as the Wi-Fi alliance, then irrespective of the physical location of the adjacent carrier or the variation in signal level from movement etc, the carrier at channel 36 should not be affected. Guard bands play a vital role in this. They not allow for possible ?sideband creep?, but also for frequency drift of the actual carrier frequency itself. [/quote]

    Bandwidth testing I did years ago shows that adjacent 5 GHz channels are still affected by sideband interference. I am planning to repeat the testing soon and put up the results, when I get a chance.

Page 1 of 2