cancel
Showing results for 
Search instead for 
Did you mean: 

BANDWIDTH FOR EACH CLIENT

f_f_5638114
New Contributor II
the bandwidth of let say 802.11g is 54mbps ,when 2 clients join the same ap and the same channel becomes 54/2 and when a third client comes it becomes (54/2)/2 and so on,how lets say the 7762 supports 256 clients ?what bandwidth 256 clients will have?
14 REPLIES 14

sid_sok
Contributor II
If price is the main concern and there are no environmental challenges then any AP will work, just keep in mind the math.

max_o_driscoll
Valued Contributor
Most of those numbers are theoretical maximums. Packets are not pure user data! They contain "other" bits to make the whole system work - all that layering stuff and IP info has to be transmitted, plus all the negotiation and re-transmission packet losses. There is also some processing overhead at client and AP that slows throughput as well.

54mbs sounds great but useful throughput at the client could be well down on that. In my experience 60-70% of nominal throughput is best you can expect.
One client to one AP at 54Mbs will not (if you download a large file) transfer at 54Mbs. More like 32-40Mbs ie 4-5MBs.

You need to allow for "inefficiencies" that are built into the system if you require a specific measurable throughput at each client.

Analogy is that wired protocol speeds are sub headline rates: fast ethernet TCP gives 90% of nominal 100Mbs. with wired gigabit giving circa 60% of nominal 1000Mbs.

These are what I see, mileage will vary with other network setups and hardware at each end, just giving real world approximations.

Anyone with with serious techie knowledge is welcome to weigh in and put me right.

david_moore_564
New Contributor
With all things being equal between APs and with the RF environment, the greatest factor affecting capacity is the capability of the clients. Although this may sound obvious, it is often forgotten. An AP that is not otherwise limited (see below) could support more 3x3:3 laptops than it can 802.11g only smartphones.

Better RF will translate to higher modulation rates (higher throughput) no matter quality of the clients. Although the more capable the client, the higher the affect. This also translates into less contention. The more quickly a device transmits, the less time it will spend on the air, leaving more air time for other devices to transmit. Anything that can be done to increase modulation rates will absolutely increase capacity. BeamFlex, to include PD-MRC, plays a major role in achieving higher modulation rates for more clients.

While the capacity of an AP is greatly affected by the RF conditions and the ability of the clients, it is also affected by the processor capacity of the AP as well. A "cheap" AP will hit a limit on client connectivity that has nothing to do with RF or the 802.11 protocol. Replacing a high end AP with two low end APs does not give you twice the capacity and can give you less.

f_f_5638114
New Contributor II
"Replacing a high end AP with two low end APs does not give you twice the capacity and can give you less"

maybe you explain what you say but i didnt understand it :)

2 different channels is 50mbps/50 clients times 2
1 channel is 50mbps/50 clients without times 1

its not like this?

david_moore_564
New Contributor
There are multiple factors that affect the capacity of an AP.
- Protocol used (802.11b vs 802.11a/g vs 802.11n)
- Channelization (20 MHz vs 40 MHz)
- RF conditions (multi-path, noise, adjacent channel and/or co-channel interference, etc)
- AP type (1x1:1 vs 2x2:2 vs 3x3:2 vs 3x3:3, etc)
- Client type (1x1:1 vs 2x2:2 vs 3x3:3, etc)
- Air time fairness (this is a big one)
- Processor power and design of the AP

I'm sure someone could come up with others. But it is the last one to which I am referring. Simply because an AP can provide 50 Mb/s of throughput does not mean that it can support 50 clients simultaneously. For any AP, there will be a point where additional clients will greatly affect performance beyond what is explained by the data transmitted. For a "cheap" AP that is designed for the home or a small office, the AP design is not typically intended to support 50 clients and performance will be limited as much or more by the number of clients than by the protocol. Don't assume that any AP can provide 50 Mb/s aggregate for 50 clients.

Now that I've opened up the can of worms a bit more and mentioned air time fairness, this deserves some discussion as well, especially since a "cheap" AP is unlikely to provide this feature. With air time fairness, each client is provided equal air time independent of the protocol used. That is, 802.11b clients, 802.11g clients, and 802.11n clients are all given equal air time as long as they all have packets to transmit. Without this, the statistical nature of WiFi is such that slower clients will take up more air time communicating with the AP. And, without air time fairness, slow clients will end up dragging down the throughput of all clients to their speed. This is mitigated somewhat by block acknowledgments for 802.11n clients, but it is still an issue. Also consider that even if you had all 802.11n clients, some will be faster than others, but all will be limited by the slowest client if you do not have air time fairness. With air time fairness, faster clients have equal time access, but will be able to transmit more frames in that amount of time than slower clients will. Faster clients will be able to achieve higher throughput than slower clients. Overall, the capacity of the AP increases.

Let's take your example of 50 clients on an AP. It is conceivable that all clients will operate at the same data rate, but in reality this is pretty unlikely. What is more realistic is to assume that you have a variety of data rates. With air time fairness, the AP can maximize capacity rather than limiting all client rates to the slowest one. Again, a "cheap" AP is unlikely to support air time fairness and will be subjected to this problem. With all clients taking longer to transmit data, they are on the air more and there is more contention, further reducing capacity for the AP. The problem builds on itself. You can see the affect of this in some of the tests performed by organizations such as WLAN Pros and Tom's Hardware. APs that don't provide air time fairness were not able to support 50 or, in some cases, even 10 concurrent clients streaming data. APs that did provide air time fairness supported the greatest number of concurrent clients.